Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Protocol - Intervention

Iron supplementation for breath‐holding attacks in children

This is not the most recent version

Collapse all Expand all

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To assess the effect of iron supplementation on the frequency and severity of breath‐holding attacks in children.

Background

Description of the condition

"There is a disease. . . in children from anger or grief, when the spirits are much stirred and run from the heart to the diaphragms forceably, and hinder or stop the breath. . . but when the passion ceaseth, this symptom ceaseth" (Culpeper 1651).

Breath‐holding attacks (BHA) are paroxysmal events affecting approximately 5% of healthy children. Breath‐holding attacks are also called breath‐holding spells by some. Two clinical forms of BHA have been delineated: cyanotic and pallid. Diagnosis is based on a distinctive and stereotyped sequence of clinical events, beginning with a provocation resulting in crying or emotional upset. This leads to a noiseless state of expiration accompanied by colour change and ultimately loss of consciousness and changes in postural tone (Lombroso 1967; DiMario 1992). Severe breath‐holding attacks are defined as those attacks resulting in loss of consciousness and/or 'convulsions' (Linder 1968; DiMario 1992; DiMario 1999).

Breath‐holding attacks occur in specific circumstances, thus are reproducible, and rarely occur when seated or lying. Sympathetic nervous system activation ('fight‐or‐flight' response) may occur resulting in the child sweating and becoming pale, taking on an almost ‘deathly’ appearance (pallid breath‐holding attack) or becoming blue in the face and lips (cyanotic breath‐holding attack). Should the attack continue, loss of consciousness may result from a fall in cerebral blood flow. This may be accompanied by sinus bradycardia (slow heart rate) and a period of asystole (pause in heart beats) on the electrocardiograph (ECG). The child falls to the ground (and may later recall this). Tonic‐clonic (jerking) movements secondary to hypoxia may occur for up to 30 seconds in duration in up to 15% of children with breath‐holding attacks (DiMario 2001). The brainstem is electrically silent during the episode (Aicardi 1998). The electroencephalogram (EEG) is usually normal between episodes and anticonvulsants are neither indicated (Mocan 1999) nor effective (Stephenson 1978). The period of confusion following the event is short, typically less than 30 seconds. Breath‐holding attacks usually do not last more than one to three minutes in total and only occur when the child is awake.

The majority of children with breath‐holding attacks have multiple episodes per week, and as many as two thirds have two to five attacks per day (Lombroso 1967). Attacks often spontaneously cease without any medical treatment by 7 to 8 years of age (Tam 1997), with most remitting between 3 and 4 years of age.

Breath‐holding episodes can be confused with apnoeas (cessation of breathing) or seizures. Apnoeas can occur anytime while awake or asleep and are associated with the premature neonate.  Breath‐holding attacks only occur during wakefulness, the infant is generally older (aged from six months) and the episodes occur as a predictable response to an unpleasant stimulus or situation. Similarly, like apnoeas, epileptic seizures may occur spontaneously in any situation, including sleep. Generalised tonic‐clonic seizures usually begin with a cry or moan (breath‐holding attacks are silent), followed by stiffening of all limbs with superimposed jerking, often lasting more than one minute. There may be cyanosis (rather than pallor), dribbling of saliva and tongue biting. The period of confusion following an epileptic seizure typically lasts more than two minutes. The EEG displays widespread cortical electrical activity (Somerville 2007).

Although a clinical distinction between pallid and cyanotic breath‐holders has been made (Lombroso 1967), it has never been clear whether a different underlying mechanism exists to cause these two different appearances in colour on the child's face. Both types of attacks have been described to occur in the same family and within the same individual. A long held belief is that the clinical appearance of the breath‐holding attack reflected the underlying pathophysiological mechanism (Lombroso 1967; Gastaut 1958; Stephenson 1980). Pallid attacks reflected predominantly parasympathetic nervous system‐mediated cardiac inhibition (bradycardia), whilst cyanotic attacks were thought to be due to predominantly intense sympathetic nervous system‐mediated respiratory inhibition. Respiratory sinus arrhythmia, an indirect measure of parasympathetic (vagal) tone, and a prolonged asystole time (<1 to 2 seconds) occurs in both pallid and cyanotic breath‐holding attacks. Although these have been reported to occur more often in pallid breath‐holding attacks, the results have not been statistically significant (Kolkiran 2005). In a case report by Tam and Rash (Tam 1997), an infant is described as having cyanotic breath‐holding attacks but was noted to have pallor with loss of consciousness. In breath‐holders, various combinations of sudden apnoea, bradycardia, asystole, and cerebral ischaemia may produce various clinical “types” (pallid, cyanotic or mixed) of breath‐holding attacks.  No specific clinical sign has been found to be pathognomonic for the type of breath‐holding attack. No significant differences in clinical follow‐up data have been observed between breath‐holders with predominant cyanotic or pallid attacks. The approximate ratio of cyanotic: pallid: both has emerged as a 5:3:2 ratio fairly consistently (DiMario 2001). More importantly, the presence of iron deficiency significantly prolongs the duration of asystole during breath‐holding attacks, regardless of type (Kolkiran 2005).

Some centres report a prevalence of simple or 'benign' breath‐holding attacks (ie those attacks occurring without loss of consciousness and/or jerking movements) as high as 27% of well children (Colina 1995; Daoud 1997; Bridge 1943). This high value may be accounted for by including other behavioural phenomenon, such as temper tantrums. Severe breath‐holding attacks of either type occur in approximately 0.1% to 4.6% of healthy children (DiMario 1992). There is no gender difference. The youngest case reported in the literature is a 3‐day‐old infant with positive family history (Breukels 2002), whereas the oldest reported case involves a patient 11 years and 8 months old (Low 1955). Breath‐holding attacks are most common in children aged 6 months to 6 years, with 76% of cases occurring between 6 and 18 months of age (Zubcevic 2000). Since breath‐holding attacks occur rarely over the age of six years, an alternate diagnosis should be sought and considered in this age group before an attack is labelled as such. A prospective cohort study (DiMario 2001) recorded a weekly median frequency of breath‐holding attacks with 30% of children experiencing one or more attacks per day.

Between 20% and 30% of children have an affected family member (Lombroso 1967; DiMario 1990) suggesting a genetic contribution, with equal frequency distributed between paternal and maternal sides. DiMario and Sarfarazi demonstrated a low‐penetrance autosomal trait (DiMario 1997; DiMario 1997a), meaning that whilst breath‐holding attacks tend to occur in children with a positive family history more often than not, this is not always the case. Family structure and mother’s attitude do not seem to have an effect on the development of breath‐holding attacks (Hannon 1997).

If a treatment reduced the severity or frequency of breath‐holding attacks and in turn improved the child’s quality of life and/or reduced parental stress, this would be clinically beneficial. Issues of cost, administration and possible adverse effects need naturally to be considered.

Children who have had breath‐holding attacks have a higher incidence of syncope as adolescents than the rest of the population (Zubcevic 2000).

Description of the intervention

The pathophysiologic mechanism and treatment of breath‐holding attacks remain controversial and unclear (Colina 1995; DiMario 1992). Autonomic nervous system dysfunction appears to play a role in the development of breath‐holding attacks (DiMario 1990). Autonomic dysregulation resulting in vagally mediated cardiac arrest and subsequent cerebral anoxia has been proposed (Daoud 1997; Holowach 1963).

Anaemia has been suggested to exacerbate the likelihood of breath‐holding episodes because the lower haemoglobin results in more rapid cerebral anoxia secondary to decreased oxygen‐carrying capacity (Colina 1995; Holowach 1963). Also, iron‐deficient children are more irritable, which increases the likelihood of an attack (Colina 1995).

First reports of an association between breath‐holding episodes and anaemia came from a retrospective study (Holowach 1963). In that population the lower the haemoglobin at onset of breath‐holding episodes, the more likely the patient was to respond to iron supplementation, with several children free of attacks after the anaemia was treated (Holowach 1963). From this study it is unclear whether the iron therapy or simply the resolution of the anaemia resulted in a cessation of the breath‐holding attacks. However, one case study (Tam 1997) showed that breath‐holding attacks improve with iron supplementation before the resolution of anaemia. This suggests that iron deficiency with or without anaemia per se is implicated in breath‐holding attacks.

In 2003, UNICEF estimated that 40% to 50% of children under 5 years old in developing countries were iron deficient (United Nations Children’s Fund 2003). Children less than two years old in metropolitan developed countries have a prevalence of iron deficiency of approximately 7% (Karr 1996) with higher rates in indigenous populations. The diagnosis of iron deficiency in children is problematic with no single laboratory test having sufficient sensitivity and specificity to be used as a sole diagnostic criterion. Indeed, 8% children with no abnormal laboratory test results, 11% with one abnormal result, 28% with two abnormal results and 63% with three abnormal results will be anaemic within a population (Cook 1976). Reference ranges for red cell indices, serum ferritin and iron saturation vary with age. Intercurrent infection will affect laboratory parameters used to diagnose iron deficiency, eg reduced serum haemoglobin or raised ferritin levels. The “gold standard” for assessment of iron stores in adults, bone marrow examination, is less useful in children, given that the majority of normal children have minimal iron stores until adolescence. Erythrocyte indices such as hypochromicity, microcytosis and reticulocyte count with treatment offer more practical investigations (Grant 2007). Soluble transferrin receptor levels may improve the diagnosis of anaemia in settings with a high burden of infectious diseases and iron deficiency (Richie 2004). Alternatively, an increase in haemoglobin of more than 10 g/L after treatment with 3mg/kg/day of elemental iron for one month is diagnostic of iron deficiency (Oski 1993).

Two prospective studies (Donma 1998; Azam 2008) of piracetam, a cyclic derivative drug of gamma‐aminobutyric acid (GABA) and pharmacologically unrelated to iron, suggest this may be an effective prophylactic treatment for severe breath‐holding attacks in children without anaemia (Garg 1998). However, in comparison to iron, this drug is relatively new, more expensive, has little data for use in children, may cause side‐effects such as agitation and requires cautious use in the presence of renal impairment. For these reasons and until more is known about the drug, it will not be currently considered in this review.

Implantable cardiac pacemakers have been used to treat some cases of severe, prolonged breath‐holding attacks associated with life‐threatening bradycardia or asystole (Kelly 2001). This small select group of children aged 1‐5 years, had been trialled on anticonvulsant, anticholinergic and theophylline medication to try to prevent breath‐holding attacks, all without success. (Iron was not used). The pacemakers reduced the severity of the attacks but the onset and/or frequency of the breath‐holding attacks were not completely prevented. If a simple, readily obtainable, inexpensive and non‐invasive treatment option for breath‐holding attacks were available, clearly this would be preferable. Oral iron supplementation fulfils these criteria and has been suggested to reduce the onset and frequency of breath‐holding attacks (which were not even completely prevented by implantable cardiac pacemakers), making it an exciting treatment modality worthy of further study.

How the intervention might work

It is not known whether, or how, iron deficiency leads to breath‐holding attacks. Iron deficiency anaemia may lead to adverse effects on oxygen uptake in the lungs and reduce available oxygen to the tissues, including central nervous system tissue (Samuels 1991; Poets 1992). Not all children with breath‐holding attacks are iron deficient at baseline (Daoud 1997; Mocan 1999). The majority of studies classify children as being iron deficient on the basis of low haematological indices, such as haemoglobin, serum transferrin, mean corpuscular haemoglobin (MCH) and mean corpuscular volume (MCV), without any evidence of bone marrow failure. Values below the normal range signify iron deficiency anaemia. It is unclear why iron‐replete children may respond to iron therapy.

It is possible that iron supplementation maximises neurotransmitter synthesis and function.  Iron has a role in catecholamine metabolism and the functioning of enzymes and neurotransmitters in the central nervous system (Oski 1978). The resolution of attacks during treatment with iron may be related to the functional restoration of these neurotransmitters.

The role of iron in the maintenance of a functional nervous system has been studied by a variety of investigators. One study (Voorhess 1975) found increased urinary excretion of adrenaline as a result of reduced monoamine oxidase activity in patients with iron‐deficiency anaemia. Activity of aldehyde oxidase, a key enzyme in serotonin degradation, also was noted to be reduced with iron deficiency (Mackler 1978). Iron also plays an important role as a cofactor in neurotransmitter synthesis and myelination (Connor 1994). Iron‐deficiency anaemia therefore appears to affect normal neurologic function.

It is hypothesised that the clinical and haematological picture of breath‐holding attacks relates to the interactions of cerebral erythropoietin, nitric oxide and interleukin‐1 (Mocan 1998; Masuda 1994). It has been suggested that “increased brain erythropoietin production has a protective effect during breath‐holding attacks, but where this does not compensate for the severity of anoxic attacks, tonic‐clonic movements may develop.” (Mocan 1999).  In addition it has been postulated that the enhanced erythropoiesis resulting from increased EPO secretion during hypoxia can also induce at least a temporary state of iron‐deficient erythropoiesis in iron‐replete patients (Mocan 1998).

Iron deficiency also has been associated with a variety of neurologic and behavioural manifestations, including pseudotumor cerebri, papilloedema, syncope, cranial neuropathy, and decreased intellectual function (Bruggers 1990; Oski 1979). Iron deficiency with or without anaemia per se may adversely affect autonomic nervous function.

Why it is important to do this review

Breath‐holding attacks are a common ailment of childhood.  Although it is a self‐resolving condition, the attacks are stressful for the parents/carers and provoke frequent presentation to primary health care facilities and emergency departments, when associated with tonic‐clonic movements.  Iron therapy is a relatively low‐cost and freely available treatment that is believed to assist in reducing this condition without causing unacceptable adverse drug effects. Although a few trials have shown promising results, results from individual trials have varied. A systematic review of iron use for breath‐holding attacks will bring together existing evidence to assess whether there is sufficient information to recommend this therapy as an effective treatment.

Objectives

To assess the effect of iron supplementation on the frequency and severity of breath‐holding attacks in children.

Methods

Criteria for considering studies for this review

Types of studies

Randomised and quasi‐randomised controlled trials

Types of participants

Children aged 0 to 18 years with recurrent (>3) episodes of breath‐holding reported by an observer (usually a parent, care‐giver or clinician) 

Types of interventions

Iron supplementation versus no therapy or placebo.

Types of outcome measures

Primary outcomes

Reduction in the frequency (number over time) and/or severity (cessation of loss of consciousness and/or convulsions) of breath‐holding attacks

Timing of outcome assessment

If data are available, outcomes will be examined post‐intervention (immediately up to one month), as short‐term (one to three months), medium term (three to six months) and long term (6‐12 months or longer).

Secondary outcomes

Adverse effects of intervention (eg gastrointestinal upset and dark stools with oral iron therapy)

Number of admissions/ presentations for medical care

Days of school missed

Days of work missed (parents)

Quality of life measures (using a standardised and validated instrument)

Parental stress/attitudes (reported on a standardised and/or validated measurement scale)

Costs

Search methods for identification of studies

Electronic searches

The Cochrane Developmental, Psychosocial and Learning Problems Group  Trials register will be searched. The Cochrane Central Register of Controlled Trials (CENTRAL) will be searched using the following terms:

Search Strategy:

1     breath holding.tw.

2     (breath adj3 hold$).tw.

3     Apnea/

4     or/1‐3

5     Iron/

6     iron.tw.

7     or/5‐6

8     4 and 7

9     adolescent/ or child/ or child, preschool/ or infant/

10     (baby or babies or infant$ or toddler$ or preschool$ or pre‐school$ or schoolchild$ or child$ or boy$ or girl$ or teen$ or adolescen$).tw.

11     10 or 9

12     8 and 11

The following databases will also be searched:  MEDLINE, EMBASE, PsycINFO, CINAHL and the metaRegister of Controlled Trials.

No language or date restriction will be applied to the searches. No RCT filter will be used as it is felt that the inclusion of one may lead to relevant references being missed.

Reference lists of all reviews or summary articles identified by our search will be examined to identify any other trials.

Two of the authors (NO and AZ) will write to known content experts and to pharmaceutical companies involved in producing iron supplementation products for use in children to see if any unpublished studies relating to the use of iron supplementation for breath‐holding attacks exist.

Data collection and analysis

Selection of studies

The titles and abstracts of all studies identified by the search strategy will be reviewed independently by two authors (AZ and NO) to determine suitability for inclusion.  Where there is insufficient information to make a decision, the full text of the articles will be reviewed.  Disagreements will be resolved by consensus and arbitration by a third author (AB or KW). Justification for exclusion of studies will be documented.

Data extraction and management

Data from included studies will be extracted independently by two review authors (AZ and NO) using a pre‐determined standardised form. The following data will be extracted:

Characteristics of trials ‐ publication status, year, country of study, funding, setting, design, inclusion and exclusion criteria, recruitment, methods, analysis and results

Study methods ‐ method of allocation, blinding and losses after randomisation (follow up losses and drop‐outs)

Characteristics of participants ‐ study population, number of participants in each treatment group, age, gender, nationality and diagnostic criteria

Characteristics of interventions ‐ preparation used, dose, length of treatment and follow up, compliance, co‐interventions and intervention used in control group

Outcomes ‐ symptom score, change in clinical, laboratory or radiological findings, complications and adverse events, drop‐outs, etc.

Length of follow‐up

Assessment of risk of bias in included studies

Two authors (AZ and NO) will independently assess each included study using the risk of bias criteria outlined in chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008) without blinding to authorship or source. The assessments will be compared for inconsistencies and differences in interpretation will be resolved by discussion and consensus, with a third author (DW). Risk of bias will be assessed according to the following five domains with ratings of 'Yes' (low risk of bias); 'No' (high risk of bias) and 'Unclear' (uncertain risk of bias):

1.      Sequence generation

Was the allocation sequence adequately generated?

“Yes” (computer generated random numbers, table of random numbers, coin‐tossing or similar), “No” (day of week, even/odd clinic record number, clinician judgment, participant preference, laboratory test result such as haemoglobin value, or similar), or “Unclear” (insufficient information about the sequence generation process to permit judgment);

2.      Allocation concealment

Was allocation adequately concealed?

“Yes” (central independent unit, sequentially numbered drug containers or sealed envelopes of identical appearance, or similar), “No” (alternation or rotation, date of birth, non‐opaque envelopes, open table of random numbers or similar), or “Unclear” (randomisation stated but no information on method used is available);

3.      Blinding

Was knowledge of the allocated intervention adequately prevented during the study?

“Yes” (identical placebo medication or similar), “No” (tablets versus liquid or similar), or “Unclear” (blinding stated but no information on method used is available);

4.      Incomplete outcome data

Were incomplete data dealt with adequately by the researchers?

“Yes” (no missing outcome data, missing outcome data balanced in numbers across intervention groups and reasons for dropouts and withdrawals described or similar), “No” (reason for missing outcome data likely to be related to true outcome or similar), or “Unclear” (number or reasons for dropouts and withdrawals not described);

5.      Selective outcome reporting

Are reports of the study free of suggestion of selective outcome reporting?

“Yes” (study protocol is available, published reports include all expected outcomes or similar), “No” (not all of the study's pre‐specified primary outcomes have been reported, one or more reported primary outcomes were not pre‐specified or similar), or “Unclear” (insufficient information to permit judgement of 'adequate' or 'inadequate'); and

Any other potential sources of bias (stopping the study early, extreme baseline imbalance or some other problem) will also be explored.

Measures of treatment effect

Binary data:

For dichotomous outcomes (complete cessation of breath‐holding attacks or nil effect at all) we will record the number of participants and incidence of events in each group of the trial. The risk of the outcome for the intervention group compared to the risk for the control group will be expressed as relative risks (RR) with 95% confidence intervals (CI).

Continuous data:

Where continuous scales of measurement are used to assess the effects of treatment (such as number and/or 'severity' of breath‐holding episodes) comparisons will be made between the means of these scores. Where possible, mean difference (MD) will be calculated as the summary statistic in meta‐analyses, with standardised mean difference (SMD) used if different scales/measures have been used for the same outcome construct. Either final levels or change in levels will be included in meta‐analyses of continuous scales of measurement. When both measures are provided in a study, final levels will be included unless there is an important baseline difference between groups with regard to the outcome measure of interest. Where some studies report data as change and others only report final data with SDs, these will be plotted in separate subgroups in meta‐analyses.  When both measures are provided in a study, final levels will be included. In order to preserve the effects of randomizations and obtain the practical impact of a treatment, we will perform intention‐to‐treat analyses.

Unit of analysis issues

All included trials will be assessed to determine unit of randomisation and whether this unit of randomisation is consistent with the unit of analysis. Where cluster randomisation has occurred, an interclass correlation co‐efficient (ICC) with be extracted and used to modify the results.  Where no ICC is given and a unit of analysis error appear to exist, trial authors will be contacted and asked to provide an ICC or raw data to enable calculation of an ICC.  Where no ICC is made available, we will search for similar studies from which an ICC could be imputed and sensitivity analysis using a range of ICCs will be used to assess the impact on treatment effect.

If any cross‐over trials are found, mean and standard error of paired t‐tests will be extracted and included in meta‐analyses using the generic inverse variance function.

Dealing with missing data

Where possible, authors of the included studies will be contacted to obtain any missing data.  Where missing data cannot be obtained, we will attempt a “worst‐case/best‐case” sensitivity analysis.

Assessment of heterogeneity

Clinical heterogeneity will be assessed by the review authors, based on their agreement that study participants, interventions and outcomes are sufficiently similar to warrant meta‐analysis. Statistical heterogeneity will be assessed using the I2 statistic. A value of less than 25% will be considered low heterogeneity. Up to 75% will be considered moderate heterogeneity. A value of greater than 75% will be considered high heterogeneity. Meta‐analysis will be considered when studies are 'clinically similar' (for example, using comparable though dissimilar outcome scales, such as iron status stratification; similar though non‐identical iron doses and follow‐up period, etc), even with a high I2.

Assessment of reporting biases

If sufficient studies are found, funnel plots will be drawn to investigate any relationship between effect size and study precision (closely related to sample size). Such a relationship could be due to publication or related biases or due to systematic differences between small and large studies. If a relationship is identified clinical diversity of the studies will be further examined as a possible explanation (see also Egger 1997). Every attempt will be made to obtain unpublished data and data from conference proceedings.

Data synthesis

Assuming two or more studies that are suitable for inclusion are found, and that the studies are considered to be homogenous, a meta‐analysis will be performed on the results. Both fixed and random effects analysis will be performed as part of a sensitivity analysis.

Results of studies where the intervention is compared to different ‘comparison groups’ (eg placebo, other drug intervention) will be presented separately.

Where measures are on different scales, but those scales are clinically homogeneous, meta‐analyses will use standardised mean difference.

Subgroup analysis and investigation of heterogeneity

Subgroup analysis will be undertaken if clinically different interventions are identified or there are clinically relevant differences between participant groups.

These clinical differences may include: 

Dosage of intervention (iron)

Iron deficient or replete populations

Age of participants

Gender

Sensitivity analysis

Sensitivity analysis will be conducted to assess the impact of study quality on the results of meta‐analyses. For example, we will test to see if studies with high rates of loss to follow‐up or inadequate blinding are more likely to show positive outcomes.