Intravenous immunoglobulin for chronic inflammatory demyelinating polyradiculoneuropathy

Abstract

Background

Chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) causes progressive or relapsing weakness and numbness of the limbs, which lasts for at least two months. Uncontrolled studies have suggested that intravenous immunoglobulin (IVIg) could help to reduce symptoms. This is an update of a review first published in 2002 and last updated in 2013.

Objectives

To assess the efficacy and safety of intravenous immunoglobulin in people with chronic inflammatory demyelinating polyradiculoneuropathy.

Search methods

We searched the Cochrane Neuromuscular Specialised Register, CENTRAL, MEDLINE, Embase, and two trials registers on 8 March 2023.

Selection criteria

We selected randomised controlled trials (RCTs) and quasi‐RCTs that tested any dose of IVIg versus placebo, plasma exchange, or corticosteroids in people with definite or probable CIDP.

Data collection and analysis

We used standard Cochrane methods. Our primary outcome was significant improvement in disability within six weeks after the start of treatment, as determined and defined by the study authors. Our secondary outcomes were change in mean disability score within six weeks, change in muscle strength (Medical Research Council (MRC) sum score) within six weeks, change in mean disability score at 24 weeks or later, frequency of serious adverse events, and frequency of any adverse events. We used GRADE to assess the certainty of evidence for our main outcomes.

Main results

We included nine RCTs with 372 participants (235 male) from Europe, North America, South America, and Israel. There was low statistical heterogeneity between the trial results, and the overall risk of bias was low for all trials that contributed data to the analysis. Five trials (235 participants) compared IVIg with placebo, one trial (20 participants) compared IVIg with plasma exchange, two trials (72 participants) compared IVIg with prednisolone, and one trial (45 participants) compared IVIg with intravenous methylprednisolone (IVMP). We included one new trial in this update, though it contributed no data to any meta‐analyses.

IVIg compared with placebo increases the probability of significant improvement in disability within six weeks of the start of treatment (risk ratio (RR) 2.40, 95% confidence interval (CI) 1.72 to 3.36; number needed to treat for an additional beneficial outcome (NNTB) 4, 95% CI 3 to 5; 5 trials, 269 participants; high‐certainty evidence). Since each trial used a different disability scale and definition of significant improvement, we were unable to evaluate the clinical relevance of the pooled effect. IVIg compared with placebo improves disability measured on the Rankin scale (0 to 6, lower is better) two to six weeks after the start of treatment (mean difference (MD) −0.26 points, 95% CI −0.48 to −0.05; 3 trials, 90 participants; high‐certainty evidence). IVIg compared with placebo probably improves disability measured on the Inflammatory Neuropathy Cause and Treatment (INCAT) scale (1 to 10, lower is better) after 24 weeks (MD 0.80 points, 95% CI 0.23 to 1.37; 1 trial, 117 participants; moderate‐certainty evidence). There is probably little or no difference between IVIg and placebo in the frequency of serious adverse events (RR 0.82, 95% CI 0.36 to 1.87; 3 trials, 315 participants; moderate‐certainty evidence).

The trial comparing IVIg with plasma exchange reported none of our main outcomes.

IVIg compared with prednisolone probably has little or no effect on the probability of significant improvement in disability four weeks after the start of treatment (RR 0.91, 95% CI 0.50 to 1.68; 1 trial, 29 participants; moderate‐certainty evidence), and little or no effect on change in mean disability measured on the Rankin scale (MD 0.21 points, 95% CI −0.19 to 0.61; 1 trial, 24 participants; moderate‐certainty evidence). There is probably little or no difference between IVIg and prednisolone in the frequency of serious adverse events (RR 0.45, 95% CI 0.04 to 4.69; 1 cross‐over trial, 32 participants; moderate‐certainty evidence).

IVIg compared with IVMP probably increases the likelihood of significant improvement in disability two weeks after starting treatment (RR 1.46, 95% CI 0.40 to 5.38; 1 trial, 45 participants; moderate‐certainty evidence). IVIg compared with IVMP probably has little or no effect on change in disability measured on the Rankin scale two weeks after the start of treatment (MD 0.24 points, 95% CI −0.15 to 0.63; 1 trial, 45 participants; moderate‐certainty evidence) or on change in mean disability measured with the Overall Neuropathy Limitation Scale (ONLS, 1 to 12, lower is better) 24 weeks after the start of treatment (MD 0.03 points, 95% CI −0.91 to 0.97; 1 trial, 45 participants; moderate‐certainty evidence). The frequency of serious adverse events may be higher with IVIg compared with IVMP (RR 4.40, 95% CI 0.22 to 86.78; 1 trial, 45 participants, moderate‐certainty evidence).

Authors' conclusions

Evidence from RCTs shows that IVIg improves disability for at least two to six weeks compared with placebo, with an NNTB of 4. During this period, IVIg probably has similar efficacy to oral prednisolone and IVMP. Further placebo‐controlled trials are unlikely to change these conclusions. In one large trial, the benefit of IVIg compared with placebo in terms of improved disability score persisted for 24 weeks. Further research is needed to assess the long‐term benefits and harms of IVIg relative to other treatments.

Plain language summary

Is intravenous immunoglobulin safe and effective for people with chronic inflammatory demyelinating polyradiculoneuropathy?

Key messages

• Intravenous immunoglobulin (IVIg) leads to improvement in disability for at least two to six weeks compared with dummy treatment. 
• During this time period, IVIg is probably equally effective as plasma‐exchange and steroids.
• One study showed that the effect of IVIg persisted for six months, but more research is needed to know whether the benefit persists for longer and what the long‐term side effects are.

What is chronic inflammatory demyelinating polyradiculoneuropathy?

Chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) is an illness in which nerves become inflamed, leading to paralysis. The likely cause is that the body attacks its own nerves.

What are possible treatments for chronic inflammatory demyelinating polyradiculoneuropathy?

CIDP usually requires long‐term treatment to prevent further disability. There is much debate about the best treatment. One choice is a medicine called immunoglobulin, which is a product made from purified antibodies from human donor blood and which is given through a drip into a vein. Steroid medicines and plasma exchange (a procedure for removing and replacing the plasma component of the blood) are known to be effective.

What did we want to find out?

We wanted to know if immunoglobulin was better than placebo (dummy treatment), plasma exchange, or steroids for improving disability in people with CIDP. We also wanted to know if immunoglobulin had any unwanted effects in this population.

What did we do?

We searched for studies that examined immunoglobulin compared with placebo, plasma exchange, or steroids in people with CIDP. We compared and summarised the results of the studies and rated our confidence in the evidence, based on factors such as study methods and sizes.

What did we find?

Nine studies involving 372 people with CIDP were eligible for this review. Five studies compared immunoglobulin with placebo, one compared immunoglobulin with plasma exchange, two compared immunoglobulin with a steroid called prednisolone, and one compared immunoglobulin with a steroid called methylprednisolone.

Main results

Evidence from five studies showed that immunoglobulin compared with placebo increases the number of people who have significantly improved disability scores after two to six weeks of treatment. For every four people who receive treatment, one experiences a significant improvement. Based on evidence from one study, we can conclude that disability scores are probably better in people treated with immunoglobulin compared with people who receive placebo after 24 weeks of treatment. The risk of serious unwanted effects of immunoglobulin is probably similar to that of placebo.

Results from two other studies showed that immunoglobulin was probably no better than steroids (prednisolone or methylprednisolone) for improving disability. There is probably little or no difference in the risk of severe side effects between immunoglobulin and prednisolone, and there may be little or no difference in the risk of severe side effects between immunoglobulin and methylprednisolone. With all treatments, fewer than one in 10 people experienced a severe side effect.

What are the limitations of the evidence?

We are confident that immunoglobulin compared with placebo increases the likelihood of improved disability scores after two to six weeks of treatment. However, we have moderate confidence in most other results, because statistical analysis suggested that the effect of immunoglobulin could be beneficial or harmful.

Each study defined improvement in its own way, and the studies used different measurement scales, so it is difficult to relate them to changes in the clinical condition of people with CIDP. Further research is needed to assess the long‐term benefits and harms of immunoglobulin compared with other treatments.

How up to date is the evidence?

This is an update of a review first published in 2002 and last updated in 2013. The most recent search for studies was in March 2023.

Summary of findings

Summary of findings 1. Intravenous immunoglobulin compared to placebo for chronic inflammatory demyelinating polyradiculoneuropathy.

Intravenous immunoglobulin compared to placebo for chronic inflammatory demyelinating polyradiculoneuropathy
Patient or population: people with CIDP Settings: tertiary care centre Intervention: IVIg Comparison: placebo
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of participants analysed (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk (placebo)	Corresponding risk (IVIg)
Significant improvement in disability within 6 weeks Assessed with: scale used in original study (different scales) Follow‐up: 2 to 6 weeks	18 per 100	44 per 100 (32 to 62)	RR 2.40 (1.72 to 3.36)	269a (5 studies)	⊕⊕⊕⊕ High	The RR corresponds to an NNTB of 4 (95% CI 3 to 5) for improvement on the disability scale. Because different studies used different scales, it is difficult to assess the clinical significance of this finding.
Change in mean disability score on Rankin scale Assessed with: Rankin scale (0–6, lower is better) Follow‐up: 2 to 6 weeks	The mean improvement in disability score on the Rankin scale in the placebo group was 0.13b	0.26 points better (0.05 points better to 0.48 points better)	—	90c (3 studies)	⊕⊕⊕⊕ High	—
Change in mean disability score at 24 weeks or later Assessed with: INCAT disability score (0–10, lower is better) Follow‐up: 24 weeks	The mean improvement in disability score on the scale used in the original study at 24 weeks was 0.3 points	0.80 better (0.23 better to 1.37 more better)	—	117 (1 study)	⊕⊕⊕ Moderated	—
Serious adverse events Assessed with: questionnaires Follow‐up: 4 to 24 weeks	8 per 100	7 per 100 (3 to 16)	RR 0.82 (0.36 to 1.87)	315e (3 studies)	⊕⊕⊕ Moderatef	—
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; CIDP: chronic inflammatory demyelinating polyradiculoneuropathy; INCAT: Inflammatory Neuropathy Cause and Treatment; IVIg: intravenous immunoglobulin; NNTB: number need to treat for an additional beneficial outcome; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^a Two studies had a cross‐over design (in total 269 treatments in 235 participants).
^b Average of placebo group means.
^c One study had a cross‐over design (in total 90 treatments in 84 participants).
^d Downgraded by one level for imprecision, as only one study is included with a relatively small number of events. 
^e One study had a cross‐over design (in total 315 treatments in 288 participants).
^f Downgraded one level for imprecision as the CI incorporates a wide range of uncertainty between a reduced and increased risk of a serious adverse event.

Summary of findings 2. Intravenous immunoglobulin compared to plasma exchange for chronic inflammatory demyelinating polyradiculoneuropathy.

Intravenous immunoglobulin compared to plasma exchange for chronic inflammatory demyelinating polyradiculoneuropathy
Patient or population: people with CIDP Settings: tertiary care centre Intervention: IVIg Comparison: plasma exchange
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of participants analysed (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk (plasma exchange)	Corresponding risk (IVIg)
Significant improvement in disability within 6 weeks	No studies reported this outcome.
Change in mean disability score on Rankin scale	No studies reported this outcome. We were unable to transform the Neuropathy Disability Score (NDS) to the Rankin score with the data available.
Change in mean disability score at 24 weeks or later	No studies reported this outcome.
Serious adverse events	See comment	See comment	Not estimable	1 trial, 20 participants, 32 treatments (cross‐over)	See comment	No serious adverse events in the IVIg group
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; CIDP: chronic inflammatory demyelinating polyradiculoneuropathy; IVIg: intravenous immunoglobulin.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Summary of findings 3. Intravenous immunoglobulin compared to prednisolone for chronic inflammatory demyelinating polyradiculoneuropathy.

Intravenous immunoglobulin compared to prednisolone for chronic inflammatory demyelinating polyradiculoneuropathy
Patient or population: patients with CIDP Settings: tertiary care centre Intervention: IVIg Comparison: prednisolone
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of participants analysed (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk (prednisolone)	Corresponding risk (IVIg)
Significant improvement in disability within 6 weeks Assessed with: INCAT disability scale (0–10, lower is better) Follow‐up: 4 weeks	62 per 100	56 per 100 (31 to 100)	RR 0.91 (0.50 to 1.68)	29 (1 study)	⊕⊕⊕⊝ Moderatea
Change in mean disability score on Rankin scale Assessed with: Rankin scale (0–6, lower is better) Follow‐up: 4 weeks	The mean change in the prednisolone group was 0.17 points worse	0.21 points worse (0.19 points better to 0.61 points worse)	—	24 (1 study)	⊕⊕⊕⊝ Moderatea
Change in mean disability score at 24 weeks or later	No studies reported this outcome.
Serious adverse events Assessed with: questionnaires Follow‐up: 4 weeks	7 per 100	3 per 100 (0 to 35)	RR 0.45 (0.04 to 4.69)	57b (1 study)	⊕⊕⊕⊝ Moderatec
*The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; CIDP: chronic inflammatory demyelinating polyradiculoneuropathy; INCAT: Inflammatory Neuropathy Cause and Treatment; IVIg: intravenous immunoglobulin; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^a Downgraded one level for imprecision, as the CI includes both improvement and no improvement.
^b Cross‐over study (32 participants, 57 treatments).
^c Downgraded one level for imprecision, as the data came from one small study and the CI incorporates a wide range of uncertainty from reduced to increased risk (one unpublished trial was not used in the analysis and does not feature in this table; Camdessanche 2014).

Summary of findings 4. Intravenous immunoglobulin compared to intravenous methylprednisolone for chronic inflammatory demyelinating polyradiculoneuropathy.

Intravenous immunoglobulin compared to intravenous methylprednisolone for chronic inflammatory demyelinating polyradiculoneuropathy
Patient or population: people with CIDP Settings: tertiary care centre Intervention: IVIg Comparison: IVMP
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of participants analysed (studies)	Certainty of the evidence (GRADE)
	Assumed risk (IVMP)	Corresponding risk (IVIg)
Significant improvement in disability within 6 weeks Assessed with: ONLS disability score (0–12, lower is better) Follow‐up: 2 weeks	14 per 100	21 per 100 (6 to 77)	RR 1.46 (0.40 to 5.38)	45 (1 study)	⊕⊕⊕⊝ Moderatea
Change in mean disability score on Rankin scale Assessed with: Rankin scale (0–6, lower is better) Follow‐up: 2 weeks	The mean improvement on the Rankin scale in the IVMP group was 0.32 points	0.24 points better (0.15 points better to 0.63 points better)		43 (1 study)	⊕⊕⊕⊝ Moderatea
Change in mean disability score at 24 weeks or later Assessed with: ONLS disability score (0–12, lower is better)	The mean improvement on the disability scale used in the original study in the IVMP group was 0.6 points	0.03 points better (0.91 points better to 0.97 points better)		45 (1 study)	⊕⊕⊕⊝ Moderatea
Serious adverse events Assessed with: questionnaires Follow‐up: 24 weeks	2 per 100b	9 per 100 (0 to 100)2	RR 4.40 (0.22 to 86.78)	45 (1 study)	⊕⊕⊕⊝ Moderatec
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; CIDP: chronic inflammatory demyelinating polyradiculoneuropathy; IVIg: intravenous immunoglobulin; IVMP: intravenous methylprednisolone; ONLS: Overall Neuropathy Limitation Scale; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^a Downgraded one level for imprecision, as the data came from one small study and the CI includes both improvement and no improvement.
^b No serious adverse events were observed in participants treated with IVMP. However, this is probably the result of small numbers of treated participants. Observational studies with IVMP in CIDP are lacking. In one large retrospective study including people with multiple sclerosis, serious adverse events were reported in 3/187 (2%) participants (Créange 2009). Based on these data and clinical practice, we estimated the risk of serious adverse events due to IVMP at 2 to 5 per 100.
c Downgraded one level for imprecision as the data came from one small study and the CI incorporates a wide range of uncertainty from reduced to considerably increased risk of a serious adverse event.

Background

Description of the condition

Chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) is characterised by progressive or relapsing motor symptoms and signs in more than one limb, developing over at least two months (Ad hoc subcom 1991; Barohn 1989; Dyck 1993; Feasby 1992). Approximately 80% of people with CIDP also have sensory symptoms in the affected limb(s), though these symptoms are usually mild. Proximal and distal parts of the limbs are usually affected symmetrically. Atrophy is less marked than weakness. Almost all people with CIDP experience loss of reflexes, though in some cases only in the ankles. There have been infrequent reports of cranial nerve involvement, sometimes preceding the neuropathy (Barohn 1989; Donaghy 1985).

The overall crude prevalence of definite and probable CIDP is 1 to 9 per 100,000 adults (Broers 2019; Laughlin 2009; Lunn 1999; McLeod 1999), and 0.5 per 100,000 children (McLeod 1999). The estimated crude annual incidence is 0.15 to 1.6 per 100,000 adults. CIDP can cause prolonged periods of disability, and 54% of people with CIDP are severely disabled (modified Rankin score of 4 or 5) at some time during the illness (Lunn 1999). More than 10% of people remain severely disabled despite treatment (Chiò 2007; Lunn 1999). CIDP can occur at any age, with a peak prevalence in the sixth and seventh decade of life (McLeod 1999).

The diagnosis of CIDP is based on clinical features, electrophysiological testing, cerebrospinal fluid characteristics, and, to a limited degree, histopathological findings (van den Bergh 2010). Cerebrospinal fluid protein levels are generally elevated without cellular reaction (Dyck 1993). Electrophysiologically, the disease is characterised by reduced nerve conduction velocities, abnormal temporal dispersion, and conduction block (Barohn 1989; van den Bergh 2010; van der Meché 1989; van Doorn 1991). Pathological findings include multifocal demyelination, mononuclear cells in close approximation to demyelinated axons, remyelination, fibre loss, and 'onion bulbs' (Barohn 1989; Dyck 1993). Lymphocytic infiltration can be found in sural nerve biopsies, but as may be expected from a predominantly motor neuropathy with multifocal involvement, biopsies are frequently normal (Matsumuro 1994; Molenaar 1998).

It is important to distinguish CIDP from Guillain‐Barré syndrome (GBS), because the course, treatment, and outcome of the two disorders differ (Pollard 1987; Simmons 1993). CIDP is often considered a chronic variant of GBS. Differentiation of the two disorders relies on arbitrary clinical criteria related to the time necessary to reach maximum deficit, which is less than four weeks in GBS and more than eight weeks in CIDP (Ad hoc subcom 1991; Feasby 1992). However, CIDP may also present acutely (acute‐onset CIDP) in up to 13% of people, with the nadir of symptoms and signs reached within four weeks (Ruts 2005; Ruts 2010). The long‐term prognosis of the two conditions also differs, as fewer than 20% of people with CIDP make a spontaneous recovery.

The differential diagnosis of CIDP should also include inherited demyelinating polyneuropathies, metabolic neuropathies (associated with diabetes, uraemia, acromegaly, hepatitis, amyloidosis, and hypothyroidism), paraneoplastic neuropathies, neuropathies associated with monoclonal gammopathies, neuropathies associated with HIV or Lyme disease, and multifocal motor neuropathy.

Description of the intervention

CIDP is an immune‐mediated disorder. Randomised clinical trials (RCTs) have shown that immunosuppressive therapy (e.g. corticosteroids) and plasma exchange have a beneficial effect in people with CIDP (Briellmann 1998; Dyck 1982; Dyck 1986; Gross 1981; Hahn 1998; Server 1979; van der Meché 1997). Several uncontrolled studies have suggested that intravenous immunoglobulin (IVIg) is also beneficial (Cornblath 1991; Faed 1989; van der Meché 1989; van Doorn 1990a; van Doorn 1991; Vermeulen 1985).

How the intervention might work

The mode or modes by which IVIg exerts a beneficial effect in people with CIDP are unclear, but researchers have proposed various mechanisms of improvement (van Schaik 1994; Yu 1999). Studies of IVIg in other diseases have demonstrated that the treatment may inhibit autoantibody production, neutralise pathogenic antibodies, and decrease antibody‐dependent cellular cytotoxicity by blocking Fc‐receptors on macrophages (Kazatchkine 2001). Furthermore, peripheral blood from people treated with IVIg shows increased CD8‐positive suppressor T‐cell function.

Why it is important to do this review

This is an update of a Cochrane review first published in 2002 and last updated in 2013. The aim of this update was to include new trials, identify ongoing trials, or 'close' the review (if we found no new or ongoing trials and considered it required no further updates).

Objectives

To assess the efficacy and safety of intravenous immunoglobulin in people with chronic inflammatory demyelinating polyradiculoneuropathy.

Methods

Criteria for considering studies for this review

Types of studies

We searched for all RCTs or quasi‐RCTs that examined the effects of IVIg treatment in people with CIDP. We considered quasi‐RCTs, which allocate participants to interventions using partly systematic methods (e.g. alternation or case record numbers) despite their higher risk of selection bias, because there is a paucity of evidence on this topic. Randomised cross‐over trials were eligible. We included studies regardless of publication status and language of publication.

Types of participants

Eligible studies had to include people with definite or probable CIDP, defined according to the following characteristics.

• Symptoms and signs of polyneuropathy in the absence of systemic disease

• Electrophysiological diagnosis of demyelinating polyneuropathy (based on reduced nerve conduction velocities, conduction blocks, or both, consistent with demyelination)

• Progression of weakness exceeding eight weeks

Increased cerebrospinal fluid protein (more than 0.5 g/L) without pleocytosis was considered supportive, but not essential for the diagnosis. It was necessary for participants to have normal erythrocyte sedimentation rate, haematocrit, white cell and platelet counts, serum creatinine, serum glucose, and liver and thyroid function tests; absent antinuclear antibodies, cryoglobulins, and monoclonal proteins; and a normal chest radiograph. Participants were excluded if they had mutilation of hands or feet, retinitis pigmentosa, ichthyosis, drug or toxic exposure known to cause peripheral neuropathy, a family history of demyelinating polyneuropathy, clinical suspicion of a vasculitic disorder, sensory level at examination, or unequivocal sphincter disturbances. A nerve biopsy was not mandatory. Participants specifically included in trials based on their prior IVIg treatment response were not eligible for inclusion.

We set no restrictions in terms of trial setting. However, considering the challenge of diagnosing and treating people with CIDP, the normal settings were likely to be hospitals and neuromuscular centres.

Studies including a subset of participants that matched our inclusion criteria were eligible for inclusion if they provided separate data for the subgroup of interest.

Types of interventions

We considered studies that evaluated any dose of IVIg compared with placebo, plasma exchange, or corticosteroids. The IVIg could be from any source as long as the preparation was produced according to the World Health Organization (WHO) guidelines (WHO 1982).

Types of outcome measures

Reporting of our predefined outcomes was not an inclusion criterion of this review, and we included studies irrespective of whether they reported outcome data in a usable way.

Primary outcomes

Our primary outcome was significant improvement in disability within six weeks after the start of treatment. Since different studies used different disability scales, we analysed the proportion of participants who had a significant improvement in disability, as determined and defined by the study authors. For each study, we used the strictest available criteria to define significant improvement. Where possible, we transformed disability data to the modified 6‐point Rankin disability scale, defining significant improvement as a reduction of at least 1 point (de Haan 1993; Table 5). If necessary and possible, we asked study authors for original data to enable this transformation.

1. Transformation of Functional Disability Scale to Rankin Disability Scale.

FDS	Rankina	FDS description	Rankin description
0	0	Normal	Asymptomatic
1	1	Minor symptoms capable of running	Non‐disabling symptoms which do not interfere with lifestyle
2	2	Able to walk 30 feet without assistance but unable to run	Minor disability symptoms which lead to some restriction of lifestyle but do not interfere with the patients' capacity to look after him/herself
3	3	Able to walk 30 feet with the assistance of one person, a walker or a cane	Moderate disability symptoms which significantly interfere with lifestyle or prevent totally independent existence
4	4	Unable to walk	Moderate severe disability symptoms which clearly prevent independent existence, but do not require constant attention day and night
5	5	Requires assisted ventilation	Severely disabled, totally dependent, requiring constant attention day and night
6	—	Death	—

FDS: Functional Disability Scale.

^a The original Rankin scale, not the modified Rankin scale used in the studies (where a value of 6 indicates death).

Secondary outcomes

• Change in mean disability score within six weeks after the start of treatment (compared with baseline). We pooled data from all studies that reported change in mean disability score on any scale within six weeks. Where possible, we also converted disability scores to the Rankin scale for a separate analysis.

• Change in muscle strength, expressed as the change in mean Medical Research Council (MRC) sum score, within six weeks after the start of treatment (compared with baseline)

• Change in mean disability score at 24 weeks or later after the start of treatment (compared with baseline). We planned to pool data from all studies that reported change in mean disability score on any scale at 24 weeks or later. Where possible, we also converted disability scores to the Rankin scale for a separate analysis.

• Frequency of serious adverse events

• Frequency of any adverse events

In these analyses, for clarity of interpretation, we considered the Rankin disability scale and MRC sum score to be quasi‐linear scales.

Search methods for identification of studies

Electronic searches

On 18 May 2021 and 8 March 2023, the Information Specialist searched the following databases and trials registries.

• Cochrane Neuromuscular Specialised Register via the Cochrane Register of Studies (CRS‐Web; Appendix 1)

• Cochrane Central Register of Controlled Trials (CENTRAL) via CRS‐Web (Appendix 2)

• MEDLINE via Ovid SP (1946 to 8 March 2023; Appendix 3)

• Embase via Ovid SP (1974 to 2023 Week 9; Appendix 4)

• U.S. National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; Appendix 5)

• WHO International Clinical Trials Registry Platform (ICTRP; trialsearch.who.int; Appendix 6)

There were no limitations on language, date, document type, or publication status.

Searching other resources

We searched the reference lists of the included studies and relevant reviews, textbooks, and conference proceedings. We examined any relevant retraction statements and errata for information. We contacted investigators identified as active in the field to identify unpublished or overlooked studies. Readers are invited to suggest studies, particularly in other languages, that we should consider in future updates.

Data collection and analysis

Selection of studies

Two review authors (INvS and John Winer (JW) in the original review, FE and INvS in the 2013 update, SB and FE in this update) independently reviewed the titles and abstracts obtained from literature searches to identify potentially relevant trials. We then retrieved and assessed the full‐text publications of all potentially eligible trials. Throughout the review process, the review authors were not blinded to the authors and source institution of trials. We resolved any disagreements by consensus. We collated multiple reports of the same study.

Data extraction and management

Two review authors independently extracted data using a data extraction form. We resolved any disagreements by consensus. We would have piloted the data extraction form on at least one study had we identified additional studies that provided data for this update. We planned to extract the following study characteristics

• Study design and setting

• Characteristics of participants (e.g. age, sex)

• Eligibility criteria

• Intervention details

• Outcomes assessed

• Source(s) of study funding

• Conflicts of interest of investigators

If we had identified studies with more than two intervention arms, we would have only included the interventions that met the eligibility criteria for this review.

We did not identify any potentially eligible studies in languages that required translation. If we had, we would have arranged to have them translated.

Assessment of risk of bias in included studies

In this update, following the guidance in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions Version 5.2 (Higgins 2017), we completed a risk of bias table, addressing the following domains.

• Sequence generation

• Allocation concealment

• Blinding of participants and personnel

• Blinding of outcome assessors

• Incomplete outcome data

• Selective outcome reporting

• Other sources of bias

Two review authors independently judged each study at high, low, or unclear risk of bias in each domain, resolving any disagreements by consensus.

Measures of treatment effect

For dichotomous data, such as significant improvement in disability within six weeks, we calculated the risk ratio (RR) for each study with 95% confidence intervals (CIs). To assess overall efficacy, we pooled the RR estimates. When the 95% CI of the pooled effect estimate did not cross the line of no effect, we also calculated the number needed to treat for an additional beneficial effect (NNTB) or the number needed to treat for an additional harmful event (NNTH), defined as the number of people who need to receive IVIg rather than the comparison for one additional person to experience a given beneficial effect (in the case of NNTB) or harmful event (in the case of NNTH).

For continuous data, we used mean difference (MD; i.e. between‐group difference in the mean change scores) and 95% CI when outcomes were measured on the same scale. Where different trials used different scales to measure the same outcome, we calculated a standardised mean difference (SMD; a dimensionless measure of effect for continuous data) and 95% CI. We calculated the SMD by subtracting the mean change score of the experimental group from the mean change score of the control group and dividing the result by the pooled standard deviation (SD) of the change scores of the two groups. The SMD represents the between‐group difference in mean change scores expressed as SDs. We also used the SMD in analyses with only one trial for the sake of consistency in reporting with other comparisons.

Unit of analysis issues

We analysed the results of cross‐over trials in two ways.

• We considered cross‐over trials as two consecutive parallel design trials, assuming that no carry‐over effect had occurred; this is considered a conservative analysis method (Higgins 2022a, Higgins 2022b). We included data from participants who did not cross over to the second treatment (because of improvement during the first treatment period) in analyses for the first treatment only.

• We analysed cross‐over trials making use of the generic inverse variance (GIV) approach, using the natural logarithms of the RR and standard error (Elbourne 2002; Higgins 2022a; Higgins 2022b). In the text, we reported pooled RR estimates for all studies and for parallel‐group trials only.

For cross‐over studies, we calculated correlation coefficients for all outcomes using the formula provided in section 6.5.2.8 of the Cochrane Handbook for Systematic Reviews of Interventions, when possible (Higgins 2022b). Our preferred approach was to use individual participant data, but if they were not available, we approximated paired analyses by assuming the correlation between treatment and placebo found in the other studies. We subsequently undertook a sensitivity analysis using different values of correlation to assess the impact of the assumed value. We used the GIV method to pool both the SMD and MD of individual cross‐over studies, which we analysed separately.

If multiarm studies had been available, we would have avoided double counting participants in analyses using the methods described in Chapter 23 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022a).

Dealing with missing data

We contacted authors of all the included trials to request original trial data. We used an intention‐to‐treat (ITT) analysis for the primary outcome. Based on published and unpublished data, we performed an available‐case analysis for all other outcomes, which means we did not include missing data in the analysis. We used a last‐observation‐carried‐forward approach for study participants reaching a premature endpoint, to calculate differences from baseline measurements. In case of missing SDs, we estimated them using a correlation coefficient, as described in Chapter 6.5.2.8 of the Cochrane Handbook for Systematic Reviews of Interventions (Abrams 2005; Follmann 1992; Higgins 2022b).

Assessment of heterogeneity

We used the Chi² statistic to test for heterogeneity amongst the trials in each analysis, using P < 0.1 as the significance level. In this update, we also calculated the I² statistic, using the rough guide from Chapter 10 of the Cochrane Handbook of Systematic Reviews of Interventions to interpret the results, as follows (Deeks 2022).

• 0% to 40%: might not be important.

• 30% to 60%: may represent moderate heterogeneity.

• 50% to 90%: may represent substantial heterogeneity.

• 75% to 100%: represents considerable heterogeneity.

When interpreting the I² statistic, we took into account the magnitude and direction of effects, and the strength of evidence for heterogeneity (e.g. P value from the Chi² test, or a CI for I²: uncertainty in the value of I² is substantial when the number of studies is small).

Where there was heterogeneity, we looked for an explanation in the characteristics of the trials. Where we identified substantial unexplained heterogeneity, we reported it and explored possible causes through prespecified subgroup analyses.

Assessment of reporting biases

We did not explore small‐study biases using a funnel plot, as there were too few included studies.

Data synthesis

We used a random‐effects model in RevMan Web, as this is usually a more conservative approach (RevMan Web 2022). We undertook meta‐analysis only where this was meaningful (i.e. if the treatments, participants, and the underlying clinical question were sufficiently similar for pooling to make sense). Primary analyses included all eligible studies. We converted data on disability scales used by trial authors to the Rankin scale, obtaining individual participant data to do so where necessary and possible.

Subgroup analysis and investigation of heterogeneity

We planned to analyse subgroups of interest as described in van Doorn 1991 because of their prognostic importance in previous prospective studies and trials. However, no studies provided sufficient data on subgroups to allow for such analyses.

Sensitivity analysis

We also planned to undertake a sensitivity analysis on the basis of the risk of bias of the studies (testing the effect of removing studies at high risk of bias); however, this was not possible, as all studies that contributed data to the analysis were at unclear or low risk of bias.

Summary of findings and assessment of the certainty of the evidence

In this update, we added summary of findings tables for all comparisons with the following outcomes.

• Significant improvement in disability within six weeks after the start of treatment

• Change in mean disability score on the Rankin scale within six weeks after the start of treatment

• Change in mean disability score at 24 weeks or later after the start of treatment

• Frequency of serious adverse events

We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of a body of evidence (studies that contributed data for the prespecified outcomes) following guidance in the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2022). We considered the evidence from RCTs as very high certainty to begin with, but we downgraded by one level for serious concerns related to any of the GRADE considerations. If there were very serious concerns, we downgraded by two levels. We provided reasons for downgrading in footnotes. We resolved any disagreements by discussion or by involving another review author (IvS).

Results

Description of studies

Results of the search

The previous version of this review included eight studies (Eftimov 2013). For this update, the searches returned 952 new records (555 after deduplication). We excluded 535 records based on their titles and abstracts, leaving 20 records (13 studies) for full‐text review. One study was already included in the previous version of the review (Nobile‐Orazio 2012). We included one new study (two records; Camdessanche 2014). The remaining 11 studies were ineligible. Examining reference lists yielded no additional studies. Figure 1 illustrates the study selection process in a PRISMA flow diagram.

1.

Study flow diagram for the current update.

Included studies

We included nine RCTs involving a total of 372 participants with CIDP (Camdessanche 2014; Dyck 1994; Hahn 1996; Hughes 2001; Hughes 2008; Mendell 2001; Nobile‐Orazio 2012; Thompson 1996; Vermeulen 1993). The mean age of participants ranged from 45 years to 60 years and the percentage of males ranged from 50% to 86%. Table 6 summarises the characteristics of included trials. Five RCTs (235 participants) compared IVIg with placebo, one RCT (20 participants) compared IVIg with plasma exchange (Dyck 1994), two RCTs (72 participants) compared IVIg with prednisolone (Camdessanche 2014; Hughes 2001), and one RCT (45 participants) compared IVIg with intravenous methylprednisolone (IVMP; Nobile‐Orazio 2012). Four RCTs had a parallel‐group design (Camdessanche 2014; Mendell 2001; Nobile‐Orazio 2012; Vermeulen 1993), and five had a cross‐over design. In eight trials, the loading dose of IVIg was 2 g/kg bodyweight; the duration of administration was two days in two trials (Hughes 2001; Mendell 2001), two to four days in two trials (Hughes 2008; Nobile‐Orazio 2012), five days in three trials (Hahn 1996; Thompson 1996; Vermeulen 1993), and unspecified in one trial (Camdessanche 2014). In Hughes 2008, the baseline loading dose was followed by a maintenance dose of 1 g/kg every three weeks, while in two trials, a maintenance dose of 2 g/kg was administered every four weeks (Camdessanche 2014; Nobile‐Orazio 2012). In Nobile‐Orazio 2012, participants treated with IVMP received a total of 2 g of IVMP over four consecutive days every four weeks. In the trials comparing IVIg with prednisolone, participants received a six‐week course of oral prednisolone tapering from 60 mg to 10 mg daily (Hughes 2001), and a 0.8 mg/kg tapering course over six months (Camdessanche 2014). One study administered a total of 1.8 g/kg bodyweight of IVIg over the course of six weeks and compared this treatment with plasma exchange twice weekly for three weeks, then once weekly for another three weeks (Dyck 1994). Each study used different outcome measures. The Characteristics of included studies table provides further details.

2. Study design of the nine included trials.

Study	Design, region	na	Mean age in years	% male	CIDP criteria	Baseline disability (mean ± SD)	IVIg dose, frequency, and duration	Comparator with dose, frequency, and duration	Primary outcome and assessment time point
Vermeulen 1985	Parallel‐group Europe	28	47	71	Ad hoc subcom 1991	mRS 3b	0.4 g/kg/day for 5 days	Placebo	mRS 16 to 21 days
Dyck 1994	Cross‐over North America	20	45	50	According to Dyck 1993	NDS 76 ± 34	0.4 g/kg/week for 3 weeks, then 0.2 g/kg/week for 3 weeks	Plasma exchange 2 × per week for 3 weeks, then 1 × per week for 3 weeks	NDS, NDS‐W, EPS 6 weeks
Hahn 1996	Cross‐over North America	30	52	63	Ad hoc subcom 1991	mNDS 78 ± 27	0.4 g/kg/day for 5 days	Placebo	mNDS, clinical grade 3 to 4 weeks
Thompson 1996	Cross‐over Europe	7	46	86	Ad hoc subcom 1991	Expanded MRC sum score 53 ± 14	0.4 g/kg/day for 5 days	Placebo	Ambulation index, 10‐min walk time, expanded MRC sum score, 9‐hole peg test, myometry, HMAS 2 weeks
Hughes 2001	Cross‐over Europe	32	54	66	CIDP according to well defined criteria	INCAT‐DS 3.8 ± 1.7	2.0 g/kg over 1–2 days for 6 months	Prednisolone 60 mg for 2 weeks Tapered over course of 6 months	INCAT‐DS 2 weeks
Mendell 2001	Parallel‐group North America	53	52	52	Ad hoc subcom 1991	AMS 7.2 ± 1.2	1.0 g/kg/day for 2 days	Placebo	AMS 7 weeks
Hughes 2008	Cross‐over Europe, North America, South America, Israel	117	51	66	INCAT	aINCAT‐DS 4.2 ± 1.5	2.0 g/kg over 2 to 4 days for 24 weeks 1.0 g/kg every 3 weeks for 24 weeks	Placebo	aINCAT‐DS 24 weeks
Nobile‐Orazio 2012	Parallel‐group Europe	45	60	67	INCAT	ONLS 3.5 (1–5)b	2.0 g/kg over 4 days for 6 months	IVMP 2.0 g over 4 days 6 months	Discontinuation of treatment 6 months
Camdessanche 2014	Parallel‐group Europe	40	58c	69c	Unspecified	Not reported	2.0 g/kg, unspecified number of days, monthly, for 6 months	Prednisone 0.8 mg/kg, with an unspecified tapering regimen, for 6 months	INCAT‐DS 3 months

^a Number of participants randomised.
^b Median value (with or without range).
^c Reported for 35 participants (5 participants were excluded after randomisation for unspecified reasons).

AI: ambulation index; aINCAT‐DS: adjusted inflammatory neuropathy cause and treatment‐disability score (0–10, lower is better); AMS: average muscle score; CIDP: chronic inflammatory demyelinating polyradiculoneuropathy; EPS: electrophysiological studies; HMAS: Hammersmith Motor Ability Score; INCAT: inflammatomy neuropathy cause and treatment (1–10, lower is better); IVIg: intravenous immunoglobulin; IVMP: intravenous methylprednisolone; mNDS: modified Neuropathy Disability Score (0–262: lower is better); MRC: Medical Research Council; mRS: modified Rankin Scale (0–6: lower is better); NDS: Neuropathy Disability Score (0–244: lower is better); NDS‐W: Neuropathy Disability Score‐weakness subset (0–192: lower is better); ONLS: Overall Neuropathy Limitation Scale (0–12: lower is better).

The largest trial had a randomised response‐conditional cross‐over design (Hughes 2008). Participants who did not improve with the appointed treatment received the alternate treatment during the first study period. After 24 weeks, only participants who improved during the first study period were re‐randomised for an extension phase of another 24 weeks. In both periods, participants switched to the alternate treatment in the event of deterioration on an adjusted Inflammatory Neuropathy Cause and Treatment (INCAT) disability scale. The trialists defined the primary outcome as the percentage of participants who improved and maintained improvement from baseline on the adjusted INCAT disability scale. For the purpose of this review, we analysed only the first treatment period, because the sample entering the extension phase was largely biased. The first period can be considered a parallel‐group trial, since only participants who did not improve crossed over. The trial authors provided original data at six weeks from the first study period. We used the number of participants who achieved sufficient improvement by six weeks (and thus were not crossed over by six weeks) in the meta‐analysis for the primary outcome of this review.

We obtained individual participant data from five studies (Hughes 2001; Hughes 2008; Nobile‐Orazio 2012; Thompson 1996; Vermeulen 1993). For the study comparing IVIg to IVMP, we obtained original data on disability, Rankin score, and MRC sum score at 15 days, two months, and six months (Nobile‐Orazio 2012). We used disability and Rankin scores at 15 days for the primary outcome, as the two‐month and six‐month follow‐ups were longer than the upper limit of six weeks chosen as an appropriate treatment duration for assessment of short‐term efficacy. We sent a request for individual participant data to the principal investigators of the other trials. Two investigators indicated they would supply us with the necessary information in due course, but we did not receive the data (Hahn 1996; Mendell 2001). One investigator refused our request because "privacy of patients cannot be protected", "primary investigators no longer have a meaningful input into the handling of their data or its interpretation", and "good data may be lumped with poor data" (Dyck 1994). We obtained no individual participant data for the new included trial (Camdessanche 2014).

Excluded studies

Seven studies were excluded in previous versions of this review (Baba 1996; Curro 1987; Dalakas 1996; Hankey 1994; Kubori 1999; van Doorn 1990b; Zinman 2005). For this update, we excluded a further 11 studies, 10 of which we included in the Characteristics of excluded studies table (Breiner 2019; Cornblath 2022; EUCTR2012‐001996‐34; ISRCTN13637698; Kuitwaard 2021; Kuwabara 2017; Léger 2013; Markvardsen 2017; NCT03684018; Neacsu 2016). Reasons for exclusion were related to the study design (i.e. non‐randomised or uncontrolled studies, ineligible intervention or comparator) or participants (i.e. presence of exclusion criteria, unclear diagnostic criteria). For a detailed list of reasons for exclusion per study, see the Characteristics of excluded studies table.

Ongoing studies

We identified no relevant ongoing studies. One study listed as ongoing in the previous version of this review was terminated (confirmed by corresponding author); we moved it to the Characteristics of excluded studies table.

Risk of bias in included studies

Figure 2 shows the review authors' judgements about each risk of bias domain for each trial. We judged three studies at low risk of bias in all domains (Hughes 2001; Hughes 2008; Nobile‐Orazio 2012). The largest study was in essence a parallel‐group study with a conditional cross‐over 'escape' for participants who did not improve sufficiently (Hughes 2008). It provided the longest follow‐up. The study comparing IVIg with IVMP had an adequate follow‐up period of six months during treatment, followed by six months off treatment (Nobile‐Orazio 2012). We judged one study at high risk of bias in multiple domains (Camdessanche 2014). The remaining studies were at unclear risk of bias in one domain (Mendell 2001; Vermeulen 1993) or three domains (Dyck 1994; Hahn 1996; Thompson 1996), and low risk in the remaining domains.

2.

Review authors' judgements about each risk of bias item for each included study. Green: low risk of bias; yellow: unclear risk of bias; red: high risk of bias.

Allocation

Four studies were at unclear risk of bias for random sequence generation and allocation concealment (selection bias) because there was insufficient information for us to make a judgement (Camdessanche 2014; Dyck 1994; Hahn 1996; Thompson 1996).

We considered the remaining five studies at low risk of selection bias. One did not provide electrophysiological criteria for the definition of a demyelinating polyneuropathy, but it is unlikely that this led to selection bias (Vermeulen 1993). In Mendell 2001, a block randomisation scheme by participating centres, along with an imbalance in recruitment, accounted for a discrepancy in the number of participants in the two treatment groups. In the study comparing IVIg with IVMP, there were some nonsignificant imbalances in baseline characteristics between the treatment groups (younger age, less disability, more frequent IVIg treatment prior to study and less comorbidity in the IVIg treatment group; Nobile‐Orazio 2012).

Blinding

We judged the risk of performance and detection bias to be low in eight studies and high in one (Camdessanche 2014).

Performance bias

Seven studies provided adequate information on blinding of participants, investigators, and other caregivers, if relevant (Hahn 1996; Hughes 2001; Hughes 2008; Mendell 2001; Nobile‐Orazio 2012; Thompson 1996; Vermeulen 1993). In the study comparing IVIg with plasma exchange, participants and treatment providers were not blinded, but lack of blinding was unlikely to introduce bias (Dyck 1994). In the RCT comparing IVIg with prednisolone, some unblinding of treatment providers was expected to have taken place (Hughes 2001).

Camdessanche 2014 was an open‐label study, with both participants and researchers being aware of treatment allocation.

Detection bias

Five studies provided adequate information on blinding of outcome assessors (Hahn 1996; Hughes 2001; Hughes 2008; Nobile‐Orazio 2012; Vermeulen 1993). In two studies, the trial authors did not describe blinding of outcome assessors in detail, but we considered the risk of bias due to lack of blinding of assessors to be low (Mendell 2001; Thompson 1996). Outcomes assessors were blinded in Dyck 1993, as participants were asked not to reveal treatment allocation.

The open‐label study mentioned use of a blinded INCAT scale to score outcomes, but did not mention what this means or how blinding was achieved (Camdessanche 2014).

Incomplete outcome data

We judged the risk of bias for incomplete outcome data to be low in seven studies, unclear in one (Dyck 1994), and high in one (Camdessanche 2014).

In three studies, it was unclear whether the investigators had performed ITT analysis (Camdessanche 2014; Mendell 2001; Dyck 1994). In Dyck 1994, one participant withdrew during the first treatment period and seven participants did not cross over to the second treatment period. Only those who completed both periods were included in the analysis. In Mendell 2001, three participants withdrew from the study, and the investigators could not collect their outcome data; however, the reasons for withdrawal were unlikely to be associated with disease or treatment.

One study at high risk of bias did not provide information on five participants who were excluded after randomisation (Camdessanche 2014).

Selective reporting

We judged the risk of reporting bias to be low in eight studies, and high in one (Camdessanche 2014). In the eight studies at low risk of bias, all outcomes stated in the methods section were also reported in the results section. Camdessanche 2014 did not provide baseline characteristics, and not all outcomes stated in the methods section were reported.

Other potential sources of bias

We identified other sources of bias in five studies, which we judged at unclear risk of bias in this domain (Camdessanche 2014; Hahn 1996; Mendell 2001; Thompson 1996; Vermeulen 1993).

Camdessanche 2014 provided limited information (abstract only). Hahn 1996 included people who were receiving additional treatment, and trivial confounding was likely. In addition, Hahn 1996 used change on an outcome scale to determine whether participants crossed over in the second phase of the trial. This introduces bias, as only participants who changed were included in the second phase of the trial, and the primary analysis was performed only for participants completing both treatments. In Mendell 2001, 29 participants were analysed in the IVIg group and 21 in the placebo group; this discrepancy was due to block randomisation. Thompson 1996 made no mention of additional treatment, and the time point for assessing the primary outcome was two weeks, which we considered inappropriate. Vermeulen 1993 assessed the primary outcome between 16 and 21 days, just below the cut‐off level of three weeks that we took as the lower limit of a clinically appropriate study duration (and outcome assessment time point). This study also did not provide an electrophysiological definition of demyelinating polyneuropathy, though it refers to criteria that were commonly used at the time (Ad hoc subcom 1991).

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4

Intravenous immunoglobulin versus placebo

Five trials evaluated IVIg compared with placebo (Hahn 1996; Hughes 2008; Mendell 2001; Thompson 1996; Vermeulen 1993).

Primary outcomes

Significant improvement in disability within six weeks after the start of treatment

Five RCTs (235 participants) evaluated significant improvement in disability within six weeks (Hahn 1996; Hughes 2008; Mendell 2001; Thompson 1996; Vermeulen 1993). The assessment time points were 14 days (Thompson 1996), 16 to 21 days (Vermeulen 1993), 28 days (Hahn 1996), and 42 days (Hughes 2008; Mendell 2001). In total, the studies administered 141 IVIg treatments and 128 placebo treatments. The participants included in the three parallel‐group trials received one treatment each (104 were assigned to IVIg, 94 to placebo). Of the 37 participants included in the two cross‐over trials, three did not cross over to the second treatment because of improvement during the first treatment period; all three had received IVIg and were included in the analysis of the IVIg arm (Hahn 1996). This means that 37 IVIg treatments and 34 placebo treatments were administered in the cross‐over trials.

There were improvements reported after 78/141 IVIg treatments and after 30/128 placebo treatments. IVIg compared with placebo increases the probability of significant improvement in disability within six weeks after the start of treatment (RR 2.40, 95% CI 1.72 to 3.36; high‐certainty evidence; Analysis 1.1; Table 1). The NNTB was 4 (95% CI 2 to 5). The analysis showed no heterogeneity (Chi² = 3.34, degrees of freedom (df) = 4 (P = 0.50); I² = 0%). Among the parallel‐group trials (Hughes 2008; Mendell 2001; Vermeulen 1993), there were improvements reported after 57/104 IVIg treatments and after 25/94 placebo treatments (RR 2.14, 95% CI 1.48 to 3.09; test for overall effect: Z = 4.05, P < 0.001; Analysis 1.1). The NNTB was 4 (95% CI 3 to 6), and there was no heterogeneity between trials (Chi² = 1.79, df = 2 (P = 0.41); I² = 0%).

1.1. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 1: Significant improvement in disability within 6 weeks

We used a GIV method to analyse the primary outcome for the cross‐over trials separately, to account for paired observations. The resulting pooled effect estimate was almost identical to that of Analysis 1.1.2 (RR 3.52, 95% CI 1.58 to 7.87; Analysis 1.2).

1.2. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 2: Significant improvement in disability within 6 weeks; generic inverse variance approach

Improvement of one or more points on the Rankin scale within six weeks after the start of treatment

One trial used the Rankin scale to assess disability (Vermeulen 1993). For two trials, it was possible to transform the disability scale to a six‐point Rankin score (Mendell 2001; Thompson 1996). Mendell 2001 used a functional disability score designed by Hughes (Guillain‐Barré Syndrome Study Group 1985), which we transformed for the available cases only (Table 5). Thompson 1996 provided us with individual participant data and scored their participants on the Rankin disability scale. We were unable to transform the disability outcome to the Rankin scale with the data available in Hahn 1996 and Hughes 2008. Therefore, we analysed three RCTs (84 participants) for this outcome. Because six participants were in a cross‐over trial, the total number of treatments was 90.

The Rankin scale ranges from 0 (no symptoms) to 6 (death). In the IVIg group, 16/50 participants had an improvement of one or more points on the Rankin scale, compared with 5/40 participants in the placebo group. The probability of improvement was higher in the IVIg group, but the 95% CI included no effect (RR 2.40, 95% CI 0.98 to 5.83; Analysis 1.3). The NNTB was 6 (95% CI 3 to 50) and there was no heterogeneity between the studies (Chi² = 1.74, df = 2 (P = 0.42); I² = 0%). In the two parallel‐group RCTs (Mendell 2001; Vermeulen 1993), 15/44 participants in the IVIg group and 5/34 participants in the placebo group experienced significant improvement on the Rankin scale (RR 2.34, 95% CI 0.92 to 5.94; Analysis 1.3). The NNTB was 6 (95% CI 3 to 100). Although the I² statistic suggested moderate heterogeneity, the P value of the Chi² test was nonsignificant; the reliability of statistical tests for heterogeneity is low in analyses with few studies (Chi² = 1.69, df = 1 (P = 0.19); I² = 41%). We did not perform a GIV analysis for this outcome as only one cross‐over study (6 participants) was available for analysis.

1.3. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 3: Improvement ≥ 1 point on Rankin scale

Secondary outcomes

All secondary outcomes except serious or any adverse events were assessed at the following time points: 14 days (Thompson 1996), 16 to 21 days (Vermeulen 1993), 28 days (Hahn 1996), and 42 days (Hughes 2008; Mendell 2001).

Change in mean disability score within six weeks after the start of treatment

Five trials (215 participants) reported change in mean disability score within six weeks after the start of treatment (Hahn 1996; Hughes 2008; Mendell 2001; Thompson 1996; Vermeulen 1993). Hahn 1996 used a response‐conditional cross‐over design. To reduce bias, we decided to only include the data from the first phase for this outcome. We used the SD from the IVIg group to impute the missing SD in the placebo group. In the last update of this review, we used a correlation coefficient to approximate the missing SD in the placebo group (Eftimov 2013); this is also considered a conservative estimation. The effect estimate of the cross‐over studies changed from SMD 2.87 (95% CI −1.44 to 7.19) in the last update to SMD 1.18 (95% CI 0.37 to 1.99) in the current version. The effect estimate of all trials changed from SMD 1.37 (95% CI 0.22 to 2.53) in the last update to SMD 0.69 (95% CI 0.33 to 1.04) in the current version.

Four studies (197 treatments) used disability scales with maximum scores ranging between 5 and 14 points (Hughes 2008; Mendell 2001; Thompson 1996; Vermeulen 1993). The MD between treatments in these studies ranged from 0.17 points to 1.15 points in favour of IVIg versus placebo. Hahn 1996 used a different disability scale, and the MD between treatments in change in disability was 35.6 in favour of IVIg with a mean baseline score of 78.3 in the IVIg group. The meta‐analysis of all studies suggests that IVIg compared with placebo improves mean disability score within six weeks (SMD 0.69, 95% CI 0.33 to 1.04; Analysis 1.4). Statistical tests indicated low heterogeneity. Although there were too few studies for a reliable interpretation, the directions and sizes of effects in the analysis were fairly consistent (Chi² = 5.85, df = 4 (P = 0.21); I² = 32%). It is reasonable to expect some heterogeneity, since the studies used various scales with different numeric intervals and maximal and minimal scores to assess disability. In a sensitivity analysis, removing Hahn 1996 from the main analysis showed that this study accounted for any heterogeneity. Because the studies used different disability scales, it is difficult to translate our findings to clinical practice. Accounting for paired data in an analysis of Thompson 1996 made little difference to the SMD (Analysis 1.5).

1.4. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 4: Change in mean disability score within 6 weeks

1.5. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 5: Change in mean disability score within 6 weeks; generic inverse variance approach

Change in mean disability score on the Rankin scale within six weeks after the start of treatment

The Rankin scale ranges from 0 (no symptoms) to 6 (death). In the three trials that we were able to include in an analysis of change in Rankin score, the mean improvement ranged from 0.17 points to 0.40 points in the IVIg groups and from 0.00 points to 0.23 points in the placebo groups (Mendell 2001; Thompson 1996; Vermeulen 1993). Meta‐analysis indicated that IVIg compared with placebo improves disability according to Rankin score within six weeks of the start of treatment (MD 0.26, 95% CI −0.48 to −0.05; 3 trials, 84 participants (90 observations); high‐certainty evidence; Analysis 1.6; Table 1). There was low heterogeneity between the studies (Chi² = 2.30, df = 2 (P = 0.32); I² = 13%). We did not perform GIV analysis, as only one cross‐over study (6 participants) was available.

1.6. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 6: Change in mean disability score on Rankin scale within 6 weeks

Change in muscle strength (Medical Research Council sum score) within six weeks after the start of treatment

Two trials (35 participants) reported MRC sum scores (Thompson 1996; Vermeulen 1993). Seven participants received both IVIg and placebo in a cross‐over trial; 28 participants received either IVIg (15 participants) or placebo (13 participants) in a parallel‐group trial. Mean MRC sum scores were available for 42 treatments. Thompson 1996 used an expanded version of the MRC sum score, with the addition of the first dorsal interosseus muscle, so that the maximum score was 70. Vermeulen 1993 used the standard MRC sum score, with a total score ranging from 0 to 60. The pooled effect estimate suggested little or no difference between IVIg and placebo in terms of their effect on MRC sum score (MD 0.78, 95% CI −1.04 to 2.60; 2 trials, 35 participants (42 observations); Analysis 1.7). Although the trials used a slightly different outcome scale, there was low heterogeneity between the results (Chi² = 0.37, df = 1 (P = 0.54); I² = 0%). We did not perform a GIV analysis, as only one cross‐over study (7 participants) was available.

1.7. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 7: Change in muscle strength (mean Medical Research Council sum score) within 6 weeks

Change in mean disability score at 24 weeks or later after the start of treatment

Hughes 2008 assessed mean disability at 24 weeks on an adjusted INCAT disability score (0 to 10; with higher scores indicating more severe disability). During the first treatment period, 23/59 participants assigned to IVIg treatment did not respond and were switched to placebo, while 45/58 participants assigned to placebo did not respond and were switched to IVIg. Mean change in baseline disability was 1.1 points (SD 1.8) in the IVIg group and 0.3 points (SD 1.3) in the placebo group. IVIg compared to placebo probably improves disability according to INCAT score at 24 weeks (MD 0.80, 95% CI 0.23 to 1.37; 1 trial, 117 participants; moderate‐certainty evidence; Analysis 1.8).

1.8. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 8: Change in mean disability score at 24 weeks or later

Frequency of serious adverse events

Three trials reported serious adverse events (Hahn 1996; Hughes 2008; Mendell 2001). In Hughes 2008, which was the largest trial, 6/113 participants (5%) in the IVIg group had serious adverse events (9/1096 infusions, 0.8%), compared with 8/95 participants (8%) in the placebo group (11/595 infusions, 1.9%) over 48 weeks (two treatment periods). In the cross‐over trial including 30 participants, one participant on IVIg developed adverse events resembling aseptic meningitis, and no other adverse events were noted (Hahn 1996). In the parallel‐group trial including 50 participants (IVIg n = 29; placebo n = 21), headache, nausea, chills, fever and transient hypertension were reported in both treatment groups, although more often in the IVIg group (Mendell 2001).

There is probably little or no difference between IVIg and placebo in terms of the risk of severe adverse events (RR 0.82, 95% CI 0.36 to 1.87; 3 trials, 315 observations; moderate‐certainty evidence; Analysis 1.9; Table 1).

1.9. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 9: Frequency of serious side effects

Frequency of any adverse events

Three trials recorded any adverse events (Hahn 1996; Hughes 2008; Mendell 2001). In the largest trial, 62/113 participants in the IVIg group and 16/95 participants in the placebo group reported adverse events (Hughes 2008). The most common adverse events in the IVIg group were headache (4% of infusions) and pyrexia (2.4%).

The meta‐analysis suggested that IVIg compared with placebo had a greater risk of adverse events (RR 2.62, 95% CI 1.81 to 3.78; 3 trials, 283 participants (308 observations; Analysis 1.10). The NNTH was 4 (95% CI 3 to 5).

1.10. Analysis.

Comparison 1: Intravenous immunoglobulin (IVIg) versus placebo, Outcome 10: Frequency of any side effects

Intravenous immunoglobulin versus plasma exchange

One cross‐over trial compared IVIg with plasma exchange (Dyck 1994). Seventeen participants completed the trial; 15 received IVIg and 17 received plasma exchange.

Primary outcomes

Significant improvement in disability within six weeks after the start of treatment

We were unable to analyse our primary outcome because the publication did not report the proportion of treatment responders.

Secondary outcomes

Change in mean disability score within six weeks after the start of treatment

The mean improvement at six weeks on the Neuropathy Disability Scale (NDS; 0 to 280, lower is better) was 36.1 points (SD 32.0) after IVIg treatment and 38.3 points (SD 34.6) after plasma exchange. During the wash‐out period, participants worsened on average by 37.8 points (SD 44.8). The SMD between NDS scores after IVIg and after wash‐out was 1.92 (95% CI 1.03 to 2.82), and the SMD between NDS scores after plasma exchange and after wash‐out was 1.92 (95% CI 1.02 to 2.82). Out analysis indicated little or no difference between IVIg and plasma exchange in terms of their effect on disability according to NDS score within six weeks (SMD −0.06, 95% CI −0.76 to 0.63; Analysis 2.1).

2.1. Analysis.

Comparison 2: Intravenous immunoglobulin (IVIg) versus plasma exchange (PE), Outcome 1: Change in mean disability score within 6 weeks

Change in mean disability score on the Rankin scale within six weeks after the start of treatment

We were unable to convert the NDS to the Rankin scale with the data presented in the publication.

Change in muscle strength (Medical Research Council sum score) within six weeks after the start of treatment

Dyck 1994 did not report MRC sum scores.

Change in mean disability score at 24 weeks or later after the start of treatment

Dyck 1994 did not report change in mean disability score at 24 weeks or later.

Frequency of serious adverse events

No serious complications were recorded in the plasma exchange or IVIg treatment groups.

Frequency of any adverse events

Tn the plasma exchange group, there were two problems reported with catheters, and various minor adverse events such as light headedness, nausea, and rash. Dyck 1994 reported no minor complications for the IVIg group.

Additional outcomes

Dyck 1994 reported one other outcome measure: the weakness subset of the NDS. We calculated the MD and CIs for this outcome although we had not planned to. The mean improvement on the weakness subset of the NDS was 31.4 points (SD 31.5) after IVIg treatment and 33.4 points (SD 29.5) after plasma exchange. During the wash‐out period, participants worsened by an average of 33.4 points (SD 37.9). The MD between NDS after IVIg and after wash‐out was 1.81 (95% CI 0.94 to 2.68), and the MD between NDS after plasma exchange and after wash‐out was 1.92 (95% CI 1.09 to 2.75). Comparing IVIg with plasma exchange revealed little to no difference, with an MD of 0.06 (95% CI −0.76 to 0.63).

Intravenous immunoglobulin versus prednisolone

Two studies evaluated IVIg compared with prednisolone (Camdessanche 2014; Hughes 2001). Camdessanche 2014 appeared to be completed but was only available as an abstract. We were unable to derive data for our primary or secondary outcomes from the abstract, and the trialists could not provide us with individual participant data. As a result, our analyses contain data from one cross‐over trial with 32 participants (Hughes 2001). In the trial, 29/30 IVIg courses and 24/27 prednisolone courses were completed according to the protocol. Fifteen participants received prednisolone and 17 participants received IVIg in the first treatment period, and 24 participants completed both treatment arms. Outcomes were assessed at four weeks.

Primary outcomes

We were unable to extract the proportion of participants with significant improvement from the published paper of Hughes 2001, but the investigators provided us with these results for the first observation period.

Significant improvement in disability within six weeks after the start of treatment

Hughes 2001 measured disability on the INCAT scale, reporting an improvement of one grade or more after 9/16 IVIg treatments and after 8/13 prednisolone treatments. There is probably little or no difference between IVIg and prednisolone in terms of their effect on the probability of significant improvement in disability at four weeks (RR 0.91, 95% CI 0.50 to 1.68; moderate‐certainty evidence; Analysis 3.1; Table 3).

3.1. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 1: Significant improvement in disability within 6 weeks

Improvement of one or more points on the Rankin scale within six weeks after the start of treatment

Hughes 2001 reported an improvement of one point or more on the Rankin scale after 7/17 IVIg treatments and after 6/14 prednisolone treatments. Our analysis indicated little or no difference between IVIg and placebo in terms of their effect on the probability of significant improvement according to Rankin score after four weeks (RR 0.96, 95% CI 0.42 to 2.20; Analysis 3.2).

3.2. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 2: Improvement of ≥ 1 point on Rankin scale

Secondary outcomes

Change in mean disability score within six weeks after the start of treatment

Considering the 24 participants who completed the trial and were used for the primary analysis in the original paper, the mean improvement at two weeks on the INCAT disability scale (0 to 10, lower is better) was 0.71 grades (SD 1.27) after IVIg treatment and 0.58 grades (SD 0.93) after prednisolone treatment. Our analysis suggested little or no difference between IVIg and prednisolone in terms of their effect on change in disability score after four weeks (MD 0.13 grades, 95% CI −0.50 to 0.76; Analysis 3.3). Hughes 2001 also measured change in mean disability score at six weeks, when there was an improvement of 0.71 grades (SD 1.19) in the IVIg group and 0.62 grades (SD 1.53) in the prednisolone group (MD 0.09 grades, 95% CI −0.69 to 0.87).

3.3. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 3: Change in mean disability score within 6 weeks

In an analysis of the first treatment arm only (when 15 participants received IVIg and 17 participants received prednisolone), the mean improvement on the disability scale was 1.24 grades (SD 1.75) after IVIg treatment and 0.53 grades (SD 0.74) after prednisolone treatment (MD 0.71 grades, 95% CI −0.05 to 1.47).

Change in mean disability score on the Rankin scale within six weeks after the start of treatment

The Rankin scale ranges from 0 (no symptoms) to 6 (death). The mean Rankin score worsened by 0.38 points (SD 0.65) after IVIg treatment and by 0.17 points (SD 0.76) after prednisolone treatment. There is probably little or no difference between IVIg and prednisolone in terms of their effect on disability according to Rankin score (MD 0.21 points, 95% CI −0.19 to 0.61; moderate‐certainty evidence; Analysis 3.4; Table 3).

3.4. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 4: Change in mean disability score on Rankin scale within 6 weeks

Change in muscle strength (Medical Research Council sum score) within six weeks after the start of treatment

The standard MRC sum score ranges from 0 (paralysis) to 60 (normal strength). Hughes 2001 reported this measure for 46 treatments (23 IVIg, 23 prednisolone). The MRC sum score improved on average by 2.3 points (SD 3.2) after IVIg and by 1.8 points (SD 3.2) after prednisolone. Our analysis suggested little or no difference between IVIg and prednisolone in terms of their effect on muscle strength according to the MCS sum score (MD 0.50, 95% CI −1.35 to 2.35; Analysis 3.5).

3.5. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 5: Change in muscle strength (mean Medical Research Council sum score) within 6 weeks

Change in mean disability score at 24 weeks or later after the start of treatment

Hughes 2001 did not report change in mean disability score at 24 weeks or later.

Frequency of serious adverse events

A total of 30 IVIg treatments and 27 prednisolone courses were given. Serious adverse events were reported in three participants: one receiving IVIg and two receiving prednisolone. There is probably little or no difference between IVIg and prednisolone in terms of serious adverse events (RR 0.45, 95% CI 0.04 to 4.69; moderate‐certainty evidence; Analysis 3.6; Table 3).

3.6. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 6: Frequency of serious side effects

Camdessanche 2014 reported one allergic reaction in the IVIg group, leading to discontinuation of treatment. A further three serious adverse events in the IVIg group and one (unexpected) serious adverse event in the prednisone group were reported, but unspecified. The abstract provided no information on type of reaction or number of treatments received. For this reason, we decided not to include these adverse events in our analysis.

Frequency of any adverse events

Adverse events recorded were headache, indigestion, fever, rash, hypotension, urticaria and psychosis, occurring in 18/30 IVIg courses and in 11/27 prednisolone courses. Our analysis suggested little of no difference between IVIg and prednisolone in risk of any adverse events (RR 1.47, 95% CI 0.86 to 2.53; 1 study, 57 participants; Analysis 3.7).

3.7. Analysis.

Comparison 3: Intravenous immunoglobulin (IVIg) versus prednisolone, Outcome 7: Frequency of any side effects

Intravenous immunoglobulin versus intravenous methylprednisolone

One trial, including 45 participants, compared the efficacy and tolerability of IVIg versus IVMP administered every month for six months (Nobile‐Orazio 2012). The primary outcome of the trial was the difference in the proportion of participants discontinuing treatment because of inefficacy, adverse events, or intolerance. Lack of efficacy was defined as absence of improvement after two months of treatment or worsening after 15 days. Three of 24 participants (13%) discontinued IVIg, compared to 11/21 participants (52%) treated with IVMP. The most common reason for discontinuing treatment was lack of efficacy.

Primary outcomes

For the primary outcome of this review, we analysed data recorded at 15 days (data provided by study authors).

Significant improvement in disability within six weeks after the start of treatment

Nobile‐Orazio 2012 measured disability on the Overall Neuropathy Limitation Scale (ONLS; 0 to 12, lower is better). Five of 24 participants treated with IVIg had an improvement of one or more grade on the ONLS, compared with 3/21 participants treated with IVMP. IVIg compared with IVMP probably has little or no effect on the likelihood of significant improvement in disability according to ONLS score after two weeks (RR 1.46, 95% CI 0.40 to 5.38; moderate‐certainty evidence; Analysis 4.1; Table 4).

4.1. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 1: Significant improvement in disability within 6 weeks

Improvement of one or more points on the Rankin scale within six weeks after the start of treatment

There was an improvement of one grade or more on the modified Rankin scale in 4/24 participants treated with IVIg compared with 5/21 participants treated with IVMP. Our analysis suggests that IVIg compared to IVMP has little or no effect on the likelihood of significant improvement in disability according to Rankin score after two weeks (RR 0.70, 95% CI 0.22 to 2.27; Analysis 4.2).

4.2. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 2: Improvement of ≥ 1 point on Rankin scale

Secondary outcomes

Change in mean disability score within six weeks after the start of treatment

ONLS disability scores at 15 days were available for 43 participants of Nobile‐Orazio 2012 (24 in IVIg group, 19 in IVMP group). Mean improvement on the ONLS disability score was 0.12 grades (SD 1.3) after IVIg treatment compared with 0.16 grades (SD 1.1) after IVMP treatment. Our analysis suggests little or no difference between IVIg and IVMP in terms of their effect on disability according to ONLS score after two weeks (MD −0.04, 95% CI −0.76 to 0.68; Analysis 4.3).

4.3. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 3: Change in mean disability score within 6 weeks

Change in mean disability score on the Rankin scale within six weeks after the start of treatment

The Rankin scale ranges from 0 (no symptoms) to six (death). Rankin scores at 15 days were available for 43 participants. Mean improvement on the Rankin scale was 0.08 (SD 0.72) in the IVIg group and 0.32 (SD 0.58) in the IVMP group. There is probably little or no difference between IVIg and IVMP in terms of their effect on disability according to Rankin score after two weeks (MD 0.24, 95% CI −0.15 to 0.63; moderate‐certainty evidence; Analysis 4.4; Table 4).

4.4. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 4: Change in mean disability score on Rankin scale within 6 weeks

Change in muscle strength (Medical Research Council sum score) within six weeks after the start of treatment

Mean MRC sum scores (0 to 60, higher is better) at 15 days were available for 43 participants in Nobile‐Orazio 2012. The mean MRC sum score improved by 1.63 (SD 4.5) in the IVIg group and by 1.26 (SD 3.5) after IVMP. Our analysis suggested little or no difference between IVIg and IVMP in terms of their effect on change in muscle strength (MRC sum score) after two weeks (MD 0.37, 95% CI −2.00 to 2.74; Analysis 4.5).

4.5. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 5: Change in muscle strength (Medical Research Council sum score) within 6 weeks

Change in mean disability score at 24 weeks or later after the start of treatment

Nobile‐Orazio 2012 provided ONLS disability scores (0 to 12, lower is better) measured at 24 weeks. The mean change in ONLS score was 0.63 (SD 1.61) after IVIg treatment and 0.60 (SD 1.57) after IVMP treatment. There is probably little or no difference between IVIg and IVMP in terms of their effect on disability according to ONLS score after 24 weeks (MD 0.03, 95% CI −0.91 to 0.97; moderate‐certainty evidence; Analysis 4.6). Eleven of 21 participants treated with IVIg who completed the 24‐week period had an improvement of one grade or more on the ONLS disability scale, compared with five of nine participants treated with IVMP (data were missing for one participant treated with IVMP).

4.6. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 6: Change in mean disability score at 24 weeks or later

Frequency of serious adverse events

During the six months' follow‐up, a total of 131 IVIg treatments were administered in 24 participants, compared to 82 IVMP treatments in 21 participants. Two serious adverse events were reported in the IVIg group compared to none in the IVMP group. There may be little or no difference between IVIg and IVMP in the risk of serious adverse events (RR 4.4, 95% CI 0.22 to 86.78; Analysis 4.7; moderate‐certainty evidence; Table 4). One participant died because of cardiac arrest one month after the last IVIg treatment, and one participant died of respiratory failure three months after the last IVIg treatment.

4.7. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 7: Frequency of serious side effects

Frequency of any adverse events

Eleven of 24 (46%) participants treated with IVIg reported at least one adverse event compared with 14/21 (67%) participants treated with IVMP. Our analysis indicates little or no difference in the risk of any adverse event with IVIg compared with IVMP (RR 0.66, 95% CI 0.39 to 1.13; Analysis 4.8).

4.8. Analysis.

Comparison 4: Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP), Outcome 8: Frequency of any side effects

Discussion

Summary of main results

Nine RCTs including a total of 372 participants with CIDP were eligible for this systematic review.

Intravenous immunoglobulin versus placebo

IVIg compared to placebo increases the probability of significant improvement in disability within six weeks after the start of treatment, with a NNTB of 4 (95% CI 3 to 5). In the parallel‐group trials only, the RR was 2.14 (95% CI 1.48 to 3.09) and the NNTB was 4 (95% CI 3 to 6). The clinical relevance of each improvement cannot be derived from this analysis, because each RCT used a different disability scale with a unique definition of a significant improvement. To overcome this problem, we attempted to convert the various disability scales to the modified Rankin scale. We found that a higher proportion of participants treated with IVIg improved 1 point on the Rankin scale compared with participants who received placebo. The NNTB was 6 (95% CI 3 to 50).

IVIg compared to placebo improves disability (according to the scores used in each study and according to the Rankin score) within six weeks. Our analysis suggested little or no difference between IVIg and placebo in terms of muscle strength (MRC sum score).

Disability scores at 24 weeks or more were only available from the largest trial (117 participants; Hughes 2008). Participants received an initial loading dose of 2 g/kg followed by a maintenance dose of 1 g/kg every three weeks. IVIg compared to placebo probably improves disability according to INCAT score after 24 weeks.

Taken together, these results show that IVIg is superior to placebo in inducing an improvement in participants with CIDP, regardless of whether this improvement is assessed by disability or impairment. However, the effect on disability was more pronounced and is, in our opinion, clinically far more relevant for the participant. A minority of participants improved spontaneously (23%), but treatment with IVIg increased their chance of improvement to 55%, with an NNTB of 4 (95% CI 3 to 5). This means that one in every four people treated with IVIg will improve. However, if 10 participants receive IVIg, three will improve due to the therapy, approximately two will improve spontaneously, and four will not improve and will need additional treatment. Long‐term results from the largest study show an almost comparable beneficial effect of IVIg treatment for up to 24 weeks, with an adjusted RR of 34% and an NNTB of 3 (95% CI 2 to 6).

Intravenous immunoglobulin versus plasma exchange

For the trial comparing IVIg with plasma exchange, we were only able to analyse change in mean disability score within six weeks, finding little or no difference between the interventions.

Intravenous immunoglobulin versus prednisolone

IVIg compared with prednisolone probably has little or no effect on the likelihood of significant improvement in disability (measured on the INCAT scale and on the Rankin scale) within six weeks after the start of treatment (Table 3). In addition, we found little or no difference in mean improvement in disability according to INCAT score and Rankin score at two weeks and six weeks. The other outcome measures revealed no clinically relevant differences between the two treatment arms. No studies reported disability scores at 24 weeks or later. One apparently completed study was only published as an abstract; we were unable to obtain usable data from the publication or from the study authors (Camdessanche 2014). This study found no difference between the groups in the number of participants who improved after three and 12 months, but these unpublished findings were subject to high risk of bias in multiple domains.

Intravenous immunoglobulin versus intravenous methylprednisolone

There are probably no important differences between monthly IVIg and IVMP treatment in the key primary and secondary outcomes of this review (Table 4). However, the trial showed that IVIg is less frequently discontinued than IVMP based on inefficacy, adverse events, or intolerance (the primary outcome of the study).

Adverse events and safety

Two trials did not mention adverse events. Headache, nausea, chills, and fever were quite common but transient adverse events with IVIg treatment. They occurred in almost half the IVIg‐treated participants, around one in five placebo‐treated participants, and almost half the prednisolone‐treated participants. Serious adverse events occurred in around 6% to 7% of participants treated with IVIg, placebo, and prednisolone. Serious adverse events were more frequent in participants treated with plasma exchange (12%).

Overall completeness and applicability of evidence

This review provides high‐certainty evidence of short‐term efficacy of IVIg compared to placebo. The direction of effect was consistent in all studies. Participants were well defined according to accepted criteria (van den Bergh 2010). Study samples reflected the general characteristics of CIDP patients, with a mean age ranging from 45 years to 60 years, and a percentage of males ranging from 50% to 86%. Where the data were available, they showed that most participants had a relatively long disease duration prior to inclusion. Whether the results of this review are applicable to people with atypical CIDP variants is unclear, as these data were not available. One study excluded people with atypical CIDP altogether (Nobile‐Orazio 2012). In some studies, many participants had received previous IVIg treatment (Dyck 1994; Hahn 1996; Hughes 2008; Nobile‐Orazio 2012); and Nobile‐Orazio 2012 excluded known non‐responders; these eligibility criteria could have introduced bias. IVIg dosing and intervals were similar in the included studies, and it remains unclear whether patients would benefit equally from more tailored dosing regimens (Lunn 2016; Rajabally 2006). There is some evidence to indicate a dose‐response benefit of IVIg, favouring higher dose (maintenance) infusions (Kubori 1999; Cornblath 2022). More frequent low‐dose maintenance IVIg dosing does not seem to improve efficacy in people stable on regular IVIg treatment (Kuitwaard 2021). Only two studies reported long‐term follow‐up (Hughes 2008; Nobile‐Orazio 2012). Various studies have reported sustained improvement on IVIg treatment for up to 52 weeks (Hughes 2008; Kuwabara 2017). It should be noted that a proportion of IVIg responders may not require long‐term treatment or may show sustained improvement on lower doses, though there is no way to reliably identify these people (Hughes 2008; Nobile‐Orazio 2015; RMC Trial Group 2009). We found no ongoing trials that could be eligible for inclusion in future updates of this review.

We excluded a single‐blind parallel‐group trial that compared two doses of IVIg to a special plasma exchange system: low‐dose IVIg (0.5 g/kg/day in two days), high‐dose IVIg (1 g/kg/day in two days), and Excorim staphylococcal protein immunoadsorption plasma exchange (Zinman 2005). The advantage of this selective adsorption of immunoglobulin is that it reduces plasma protein and fluid loss caused by conventional plasma exchange. The study was at high risk of bias; of the 20 included participants, only six from the high IVIg dose group and four from the immunoadsorption group completed the trial. We could not assess our primary outcome (significant improvement in disability within six weeks), as only two‐ and six‐month assessments were available. Furthermore, we could not extract six‐month disability data, because the trial authors used a composite of different outcomes to define clinical improvement. According to the paper, three of six participants treated with IVIg and all four participants treated with immunoadsorption improved. Considering the small number of participants, larger trials are needed to assess the efficacy and safety of immunoadsorption.

In clinical phase III trials, the primary outcome should be disability rather than impairment, as the primary question to be answered is whether a person with CIDP benefits from a particular treatment. Three studies (54 participants) used an inappropriate scale for this type of clinical trial (Dyck 1994; Hahn 1996; Thompson 1996). CIDP is a chronic disease that can follow a chronic‐progressive or relapsing‐remitting course. Outcomes were assessed at 14 days (Thompson 1996), 15 days (Nobile‐Orazio 2012), 16 to 21 days (Vermeulen 1993), 28 days (Hahn 1996), and 42 days (Mendell 2001). The authors of Hughes 2008 provided original INCAT disability scores and the percentage of participants responding to treatment at six weeks. Only Nobile‐Orazio 2012 had a longer follow‐up period than Hughes 2008, up to 48 weeks. In a chronic disease, the number of participants maintaining remission for a longer period is of major importance. Future trials should therefore use long‐term rather than short‐term outcomes.

We analysed the results of the cross‐over trials as if they had come from a parallel‐group trial, assuming that no carry‐over effect had occurred. Analysing cross‐over trials in this way leads to a conservative estimate. For dichotomous data, this approach is not statistically ideal, but was unavoidable if we wanted to assess efficacy from all available studies. One way to overcome this problem is to carry out individual participant data analysis of all studies. We were able to obtain these data from two studies. In the text of the review, we analysed the dichotomous outcome parameters separately for all trials and for the parallel‐group trials. We also provided a GIV analysis for the cross‐over trials where appropriate. In general, the effect estimates were similar to the original analyses. All cross‐over trials used a predefined set of rules to determine the length of the wash‐out period, during which participants were crossed over if they remained stable, if they did not improve sufficiently, or if they deteriorated, depending on the scale used. Participants with significant improvement had to deteriorate below a predefined level before cross‐over could take place. Some participants who did not deteriorate were not crossed over at all. Only Hughes 2001 conducted a formal statistical test to exclude the possibility of a carry‐over effect. Nevertheless, we think it unlikely that a carry‐over effect accounts for the treatment effect found in this systematic review because the wash‐out periods were sufficient. In the Effects of interventions section, we have provided separate RRs for all studies and for the parallel‐group trials only (Mendell 2001; Nobile‐Orazio 2012; Vermeulen 1993).

We also planned to analyse other subgroups of interest because of their prognostic importance in previous prospective studies and trials (van Doorn 1991). However, we were unable to do so owing to insufficient data.

With some reservations, IVIg appears to be as effective as plasma exchange, prednisolone, and IVMP. Firstly, the plasma exchange versus IVIg trial used an unusual dose and regimen of IVIg and had some major methodological shortcomings. Secondly, the prednisolone versus IVIg trial was terminated prematurely because the trial medication expiry date was reached before all the intended 40 participants could be accrued (Hughes 2001). Moreover, this trial was not designed and powered to detect equivalence between the two treatments, and the regimen of prednisolone was relatively short and not typical of clinical practice for the treatment of participants with CIDP. Finally, for the trial comparing IVIg and IVMP (Nobile‐Orazio 2012), we used outcomes at 15 days as defined by our study protocol. Although in favour of IVIg, the proportion of participants with improved disability scores did not differ significantly between both groups, which was also the case when outcomes were assessed at two months (data not shown) and six months (Analysis 4.6). Participants who did not improve within two months were switched to the alternative treatment. Studies with insufficient follow‐up might underestimate the treatment effect, especially in participants treated with corticosteroids: there is evidence that more than half of IVIg responders improve between three and six weeks (Latov 2010), while the median time to improvement in participants treated with pulsed oral dexamethasone in one trial was 17 weeks (van Schaik 2010). The dose of methylprednisolone in the study included in our review was high compared with doses used in previous studies of corticosteroids (Lopate 2005; Muley 2008; van Schaik 2010). As we found little or no difference between IVIg and IVMP in terms of improvement in disability, higher dosing of corticosteroids may not be necessary, leading only to more adverse events. This could explain the high dropout rate at 15 days in the IVMP group (Nobile‐Orazio 2012). Another problem with this study was that more participants in the IVIg group were treated with IVIg previously and were known to be responsive to this treatment. Therefore, the design of the study could have overestimated the efficacy of IVIg.

Since IVIg, plasma exchange, prednisolone, and IVMP seem to be equally effective in the treatment of people with CIDP, it is unclear which treatment should be the first choice. Costs, adverse events, duration of treatment, dependency on regular hospital visits, and ease of administration all have a bearing on such a decision. Prednisolone is cheap and easy to administer. Compared with prednisolone treatment, IVIg will be cost‐effective only when one quality‐adjusted‐life‐year (QALY) is valued at over EUR 250,000 (McCrone 2003). However, adverse events due to long‐term prednisolone treatment are not taken into account and would reduce the cost per QALY of IVIg treatment. Up to 70% of people with CIDP relapse when the dose of steroids is reduced, and some studies recommend two‐year treatment schedules (Lindenbaum 2001; Mehndiratta 2012). Another Cochrane systematic review advocated the use of immunosuppressants and other immunomodulatory agents as adjuvant therapy and discussed the merits of these additional treatments in detail (Mahdi‐Rogers 2013). Approximately 25% of people with CIDP do not need further steroid treatment because they have achieved a real clinical remission. Many serious adverse events accompany long‐term treatment with prednisolone. Half‐year treatment of corticosteroid pulses, which might reduce long‐term adverse events, might be sufficient in some people with CIDP (Nobile‐Orazio 2012; van Schaik 2010). Plasma exchange is complicated, expensive, and has no advantages over immunoglobulin treatment. Immunoglobulin treatment is expensive, usually needs a hospital setting to be administered, and must be repeated almost every month for many years to maintain the best possible functional status in people with CIDP. For those who require regular maintenance IVIg treatment, subcutaneous immunoglobulins (SCIg) may be a viable alternative (van Schaik 2018). SCIg for the treatment of CIDP is the subject of another Cochrane systematic review currently in development (Bus 2021).

Quality of the evidence

We graded the certainty of evidence as moderate to high for all outcomes except serious adverse events in the IVIg versus IVMP comparison, where data were sparse. Follow‐up in older trials was relatively short. Because the trials in the IVIg versus placebo comparison used different scales to measure disability, it was difficult to determine the clinical significance of the results. We converted disability data to the Rankin scale where possible, finding a similar effect estimate and a slightly larger number needed to treat. For each of the other comparisons, the review included only one trial in the analysis, limiting robust conclusions.

Potential biases in the review process

We could not reach the corresponding authors of one study identified for full‐text review (Neacsu 2016). The study does not appear to have been published; based on available abstract data, we considered it did not meet our eligibility criteria. One study included in this review was only published as an abstract and provided no usable data (Camdessanche 2014). We cannot rule out publication bias, at least for the prednisone comparison, though the included abstract reported no difference in disability between the groups at three and 12 months (Camdessanche 2014). For the pooled studies, heterogeneity was not considered a likely source of bias. We excluded one study that included people with diabetes and features of a demyelinating polyneuropathy (Breiner 2019). This study demonstrated no effect of IVIg compared with placebo. It is unclear whether the participants met the clinical criteria for CIDP. For Hahn 1996, we verified the data used for our primary outcome measure with the trialist.

Agreements and disagreements with other studies or reviews

The short‐term efficacy results of our meta‐analysis are in line with the findings of other large prospective studies (Léger 2013; Kuwabara 2017; van Schaik 2018). Regarding long‐term efficacy, Hughes 2008 included a 24‐week extension phase in which most participants were re‐randomised to IVIg or placebo. In the IVIg group, most participants had no relapse and remained clinically stable or improved, whereas around half of participants in the placebo group had no relapse (NNTB 3, 95% CI 3 to 8). We did not include the results of the extension phase in the analyses because this sample was largely biased. Similarly, due to treatment bias, we excluded a proof‐of‐principle study in which seven people known to respond to IVIg therapy discontinued treatment and, after deterioration, were subsequently treated with IVIg or placebo in a double‐blind, cross‐over design. All participants on IVIg improved, versus none on placebo (van Doorn 1990b).

The IVIg dose in the included studies was similar and based on the largest randomised IVIg study included in this review (Hughes 2008). One trial which we excluded from our review compared three different IVIg doses (0.25 g/kg, 1.0 g/kg, and 2.0 g/kg) in 59 people with CIDP and multifocal motor neuropathy (Kubori 1999; see the Characteristics of excluded studies table). The percentage of participants on IVIg who improved was 15% for 0.25 g/kg, 21% for 1.0 g/kg, and 60% for 2.0 g/kg, illustrating that a lower dose of IVIg may not be effective.

Time to improvement was similar in the studies included in this review, though there is evidence to suggest that previously treated patients may improve much more rapidly than untreated patients (Merkies 2019). One previous study found that almost all people on regular IVIg treatment experienced mostly transient adverse events; 33% of participants experienced headache and 4% (2 participants) developed thromboembolic complications (Kuwabara 2017). Most adverse events of IVIg are mild and transient and estimated to occur in 1% to 15% of infusions (Duhem 1994; Stiehm 1996). Rashes, chills, fever, mild hypotension or hypertension, nausea, malaise, headache, and mild arthralgias seem to be the most frequent. Occasionally, short‐lasting aseptic meningitis may occur, with no (or minimal) long‐term sequelae. The most worrisome – albeit rare – complications are severe and potentially fatal anaphylactic shock, stroke, and temporary renal impairment (relatively common in people with pre‐existing renal disease). The exact incidence of these severe adverse events is unknown, though there is evidence of a higher incidence of thromboembolic events in people receiving long‐term IVIg treatment compared to population‐based estimates (Kapoor 2020). One ongoing, postmarketing, clinical pharmacovigilance study reported an adverse reaction rate of less than 0.5% for more than 26,000 participant infusions and less than 4% in 2554 participants (Martin 2000).

In this review, mild and transient adverse events were reported in 49% of IVIg‐treated participants, a far greater proportion than has been reported in the non‐randomised literature (Duhem 1994; Léger 2013; Martin 2000; Stiehm 1996). Headache was the most frequent adverse event in the included studies. As no studies reported the cumulative percentages of all adverse events, we used the percentage of headache as an estimate of any adverse event. This figure probably underestimates the adverse events of IVIg. RCTs are unsuitable for determining the frequency of adverse events. Serious adverse events were encountered in 6% of IVIg‐treated participants, which is more in line with previously published figures. In comparison with prednisolone, there was little or no difference in the occurrence of minor or major adverse events. However, it should be noted that the number of plasma exchange‐treated participants was small compared to the number of IVIg‐treated participants, so the reported frequency of adverse events in IVIg participants will be more accurate than in plasma exchange‐treated participants. Furthermore, serious adverse events known to occur after prolonged treatment with steroids were not reported in the prednisolone trial because the regimen was very short.

We found two other reviews comparing IVIg to placebo and other interventions. Gaebel 2010 identified the same studies as we did, but included other trials that we excluded. Gaebel 2010 reported a slightly higher pooled RR for improvement in disability (RR 2.7, 95% CI 1.8 to 4.2) and a comparable MD for change in disability (MD 0.65, 95% CI 0.23 to 1.1). The differences may be due to the number of studies included in the analysis, as well as the selection of participants. Another systematic review aimed to determine the effectiveness of different treatments for CIDP (Bright 2014). It included only two of the studies in our review (Hughes 2008; Vermeulen 1993), and did not perform a meta‐analysis specifically for IVIg against comparators.

The trials included in this review used various outcome scales to assess disability in people with CIDP: two types of NDS (Dyck 1992; Hahn 1996), the Hammersmith Motor Ability Score (Scott 1982), the Hughes functional disability score (GBS study group 1985), the modified Rankin scale (de Haan 1993), and the INCAT disability score (Hughes 2001; Hughes 2008; Merkies 2000). This reflects a lack of consensus on a preferred outcome measure. Even the NDS (now known as the Neuropathy Impairment Scale), used in two trials, was modified in the later trial (Hahn 1996). A 2008 review showed great diversity in outcome measures used in immune‐mediated neuropathies (van Nes 2008). The latest update of the international CIDP guidelines offers Good Practice Points (GPP) to assess disability, impairment, and improvement criteria for both (van den Bergh 2021). These criteria will hopefully allow for better comparison of results between future trials.

Authors' conclusions

Implications for practice.

The evidence from randomised controlled trials (RCTs) in people with chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) shows that intravenous immunoglobulin (IVIg) improves disability for up to 24 weeks and possibly 48 weeks compared with placebo. For every four people treated, one will have significantly improved disability within six weeks of treatment. Short‐term use of IVIg (up to six weeks) probably has similar efficacy to oral prednisolone and pulsed intravenous methylprednisolone. It is unclear which of these treatments should be the first choice. Cost, side effects, duration of treatment, dependency on regular hospital visits, and ease of administration all have to be considered before such a decision can be made.

Implications for research.

Further trials are needed to assess the long‐term benefits of IVIg, to inform guidance for maintenance treatment, and to compare the short‐term benefits of IVIg versus plasma exchange and corticosteroids. There is a need to identify more appropriate outcome measures for trials in CIDP. As IVIg is expensive, cost‐effectiveness studies comparing different treatments are also imperative.

What's new

Date	Event	Description
22 May 2024	Amended	Minor typographical error corrected in the Plain Language Summary

History

Protocol first published: Issue 3, 1999
Review first published: Issue 2, 2002

Date	Event	Description
14 February 2024	New search has been performed	Data in analysis 1.4 changed to reflect pre‐crossover participant data. SMD changed from 1.37 (95% CI 0.22 to 2.53) to 0.69 (95% CI 0.33 to 1.04). Direction of effect was unchanged; this did not impact our conclusions. In this same analysis, we used the SD of the MD in the IVIg group to estimate the missing SD in the placebo group (previously we had used a correlation coefficient). Graphs labels for the (S)AE analyses were changed to reflect the correct direction of effect. We included one new study in the review, though we were unable to include it in the meta‐analysis. Searches updated and new terms added. Methods updated to current standards. Web of Science not searched, as previous search did not produce useful results.
14 February 2024	New citation required but conclusions have not changed	One new study included in this update.
8 August 2013	New citation required and conclusions have changed	One new study involving a new comparison included Conclusions revised accordingly
4 December 2012	New search has been performed	An updated search performed on 4 December 2012
10 November 2008	New citation required and conclusions have changed	A new randomised controlled trial was identified. The review conclusions have been updated to incorporate the results of this trial.
9 November 2008	New search has been performed	An updated search of the Cochrane Neuromuscular Disease Group Trials Register was undertaken in May 2008. Supplementary searches of MEDLINE (January 1985 to May 2008), EMBASE (January 1985 to May 2008) and ISI (January 1985 to May 2008) were also undertaken. One new trial was identified which has been included.
30 September 2008	Amended	Converted to new review format.
19 June 2008	New citation required and conclusions have changed	Substantive amendment

Appendices

Appendix 1. Cochrane Neuromuscular Specialised Register via the Cochrane Register of Studies (CRS‐Web)

1 ((inflammatory NEAR3 demyelinating)):AB,EH,EMT,KW,KY,MH,TI AND INREGISTER 140

2 (polyradiculoneuropath* or polyneuropath*):AB,EH,EMT,KW,KY,MH,TI AND INREGISTER 555

3 MESH DESCRIPTOR Polyneuropathies AND INREGISTER 69

4 MESH DESCRIPTOR Polyradiculoneuropathy AND INREGISTER 56

5 (polyneuritis or polyradiculoneuritis):AB,EH,EMT,KW,KY,MH,TI AND INREGISTER 8

6 #1 OR #2 OR #3 OR #4 OR #5 563

7 MESH DESCRIPTOR Chronic Disease AND INREGISTER 206

8 #7 or "chronic disease":ti,ab 308

9 #6 AND #8 14

10 MESH DESCRIPTOR Polyradiculoneuropathy, Chronic Inflammatory Demyelinating AND INREGISTER 34

11 (chronic NEAR3 inflammatory NEAR3 demyelinating NEAR3 polyradiculoneuropath*):AB,EH,EMT,KW,KY,MH,TI AND INREGISTER 91

12 cidp:TI,AB AND INREGISTER 130

13 #9 OR #10 OR #11 OR #12 145

14 MESH DESCRIPTOR Immunoglobulins, Intravenous EXPLODE ALL AND INREGISTER 108

15 ((intravenous NEAR immunoglobulin*) or ivig):AB,EH,EMT,KW,KY,MH,TI AND INREGISTER 285

16 #14 OR #15 285

17 #13 AND #16 109

18 INREGISTER AND 18/05/2021_TO_08/03/2023:CRSCREATED 81

19 #17 AND #18 0

Appendix 2. Cochrane Central Register of Controlled Trials (CENTRAL) via the Cochrane Register of Studies (CRS‐Web)

1 ((inflammatory NEAR3 demyelinating)):AB,EH,EMT,KW,KY,MH,TI AND CENTRAL:TARGET 407

2 (polyradiculoneuropath* or polyneuropath*):AB,EH,EMT,KW,KY,MH,TI AND CENTRAL:TARGET 1880

3 MESH DESCRIPTOR Polyneuropathies AND CENTRAL:TARGET 258

4 MESH DESCRIPTOR Polyradiculoneuropathy AND CENTRAL:TARGET 91

5 (polyneuritis or polyradiculoneuritis):AB,EH,EMT,KW,KY,MH,TI AND CENTRAL:TARGET 26

6 #1 OR #2 OR #3 OR #4 OR #5 1969

7 MESH DESCRIPTOR Chronic Disease AND CENTRAL:TARGET 15182

8 #7 or "chronic disease":ti,ab 15284

9 #6 AND #8 27

10 MESH DESCRIPTOR Polyradiculoneuropathy, Chronic Inflammatory Demyelinating AND CENTRAL:TARGET 91

11 (chronic NEAR3 inflammatory NEAR3 demyelinating NEAR3 polyradiculoneuropath*):AB,EH,EMT,KW,KY,MH,TI AND CENTRAL:TARGET 193

12 cidp:TI,AB AND CENTRAL:TARGET 284

13 #9 OR #10 OR #11 OR #12 328

14 MESH DESCRIPTOR Immunoglobulins, Intravenous EXPLODE ALL AND CENTRAL:TARGET 990

15 ((intravenous NEAR immunoglobulin*) or ivig):AB,EH,EMT,KW,KY,MH,TI AND CENTRAL:TARGET 2770

16 #14 OR #15 2770

17 #13 AND #16 226

18 18/05/2021_TO_08/03/2023:CRSINCENTRAL AND CENTRAL:TARGET 246218

19 #17 AND #18 21

Appendix 3. MEDLINE (OvidSP) search strategy

Database: Ovid MEDLINE(R) ALL <1946 to March 07, 2023>

1 ((randomized controlled trial or controlled clinical trial).pt. or (randomi?ed or placebo or randomly or trial or groups).ab. or drug therapy.fs.) not (exp animals/ not humans.sh.) (4903422)

2 ((Inflammatory adj3 Demyelinating).tw. and (Polyneuropathies/ or Polyradiculoneuropathy/ or (Polyradiculoneuropath$3 or Polyneuropath$3 or Polyneuritis or Polyradiculoneuritis).tw.) and (Chronic Disease/ or "Chronic Disease".mp.)) or Polyradiculoneuropathy, Chronic Inflammatory Demyelinating/ or ((Chronic adj3 Inflammatory adj3 Demyelinating adj3 Polyradiculoneuropath$3) or CIDP).tw. (3029)

3 exp Immunoglobulins, Intravenous/ or ((Intravenous adj1 Immunoglobulin$) or IVIg).tw. (24424)

4 1 and 2 and 3 (508)

5 limit 4 to ed=20210517‐20230307 (63)

6 limit 4 to dt=20210517‐20230307 (48)

7 5 or 6 (64)

Appendix 4. Embase (OvidSP) search strategy

Database: Embase <1974 to 2023 Week 09>

1 Randomized controlled trial/ or Controlled clinical study/ or randomization/ or intermethod comparison/ or double blind procedure/ or human experiment/ or (random$ or placebo or (open adj label) or ((double or single or doubly or singly) adj (blind or blinded or blindly)) or parallel group$1 or crossover or cross over or ((assign$ or match or matched or allocation) adj5 (alternate or group$1 or intervention$1 or patient$1 or subject$1 or participant$1)) or assigned or allocated or (controlled adj7 (study or design or trial)) or volunteer or volunteers).ti,ab. or (compare or compared or comparison or trial).ti. or ((evaluated or evaluate or evaluating or assessed or assess) and (compare or compared or comparing or comparison)).ab. (6215169)

2 (random$ adj sampl$ adj7 ("cross section$" or questionnaire$1 or survey$ or database$1)).ti,ab. not (comparative study/ or controlled study/ or randomi?ed controlled.ti,ab. or randomly assigned.ti,ab.) (9369)

3 Cross‐sectional study/ not (randomized controlled trial/ or controlled clinical study/ or controlled study/ or (randomi?ed controlled or control group$1).ti,ab.) (339478)

4 (((case adj control$) and random$) not randomi?ed controlled).ti,ab. (21255)

5 (Systematic review not (trial or study)).ti. (251694)

6 (nonrandom$ not random$).ti,ab. (18738)

7 ("Random field$" or (random cluster adj3 sampl$)).ti,ab. (4433)

8 (review.ab. and review.pt.) not trial.ti. (1092886)

9 "we searched".ab. and (review.ti. or review.pt.) (48453)

10 ("update review" or (databases adj4 searched)).ab. (60401)

11 (rat or rats or mouse or mice or swine or porcine or murine or sheep or lambs or pigs or piglets or rabbit or rabbits or cat or cats or dog or dogs or cattle or bovine or monkey or monkeys or trout or marmoset$1).ti. and animal experiment/ (1215003)

12 Animal experiment/ not (human experiment/ or human/) (2551780)

13 or/2‐12 (4266479)

14 1 not 13 (5488966)

15 limit 14 to (conference abstracts or embase) (4567698)

16 ((Inflammatory adj3 Demyelinating).tw. and (Polyneuropathies/ or Polyradiculoneuropathy/ or (Polyradiculoneuropath$3 or Polyneuropath$3 or Polyneuritis or Polyradiculoneuritis).tw.) and (Chronic Disease/ or Chronic Disease.tw.)) or Chronic Inflammatory Demyelinating Polyneuropathy/ or ((Chronic adj3 Inflammatory adj3 Demyelinating adj3 Polyradiculoneuropath$3) or CIDP).tw. (6811)

17 (exp Immunoglobulin/ and (exp Intravenous Drug Administration/ or Intravenous Drug Administration.fs.)) or ((Intravenous adj1 Immunoglobulin$) or IVIg).tw. (93397)

18 15 and 16 and 17 (486)

19 limit 18 to em=202119‐202309 (46)

Appendix 5. Clinicaltrials.gov search strategy

Advanced Search

Condition or disease: Chronic Inflammatory Demyelinating Polyneuropathy

Other terms: Randomized

Study type: Interventional Studies (Clinical Trials)

Intervention/treatment: Intravenous Immunoglobulin

First Posted: From 05/18/2021 To 03/08/2023

1 Studies found

Appendix 6. WHO trials registry (ICTRP) search strategy

Advanced Search

Chronic Inflammatory Demyelinating Polyneuropathy in the Condition

Intravenous Immunoglobulin in the Intervention

Recruitment status is ALL

Date of registration is between 01/01/2021 and 08/03/2023

0 records for 0 trials found!

Data and analyses

Comparison 1. Intravenous immunoglobulin (IVIg) versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Significant improvement in disability within 6 weeks	5	269	Risk Ratio (M‐H, Fixed, 95% CI)	2.40 [1.72, 3.36]
1.1.1 Parallel design	3	198	Risk Ratio (M‐H, Fixed, 95% CI)	2.14 [1.48, 3.09]
1.1.2 Cross‐over design	2	71	Risk Ratio (M‐H, Fixed, 95% CI)	3.56 [1.59, 7.96]
1.2 Significant improvement in disability within 6 weeks; generic inverse variance approach	2	71	RR (IV, Fixed, 95% CI)	3.52 [1.58, 7.87]
1.3 Improvement ≥ 1 point on Rankin scale	3	90	Risk Ratio (M‐H, Fixed, 95% CI)	2.40 [0.98, 5.83]
1.3.1 Parallel design	2	78	Risk Ratio (M‐H, Fixed, 95% CI)	2.34 [0.92, 5.94]
1.3.2 Cross‐over design	1	12	Risk Ratio (M‐H, Fixed, 95% CI)	3.00 [0.15, 61.74]
1.4 Change in mean disability score within 6 weeks	5	227	Std. Mean Difference (IV, Random, 95% CI)	0.69 [0.33, 1.04]
1.4.1 Parallel design	3	183	Std. Mean Difference (IV, Random, 95% CI)	0.54 [0.24, 0.84]
1.4.2 Cross‐over design	2	44	Std. Mean Difference (IV, Random, 95% CI)	1.18 [0.37, 1.99]
1.5 Change in mean disability score within 6 weeks; generic inverse variance approach	1		Std. Mean Difference (IV, Random, 95% CI)	0.70 [‐0.16, 1.56]
1.6 Change in mean disability score on Rankin scale within 6 weeks	3	90	Mean Difference (IV, Fixed, 95% CI)	‐0.26 [‐0.48, ‐0.05]
1.6.1 Parallel design	2	78	Mean Difference (IV, Fixed, 95% CI)	‐0.33 [‐0.57, ‐0.09]
1.6.2 Cross‐over design	1	12	Mean Difference (IV, Fixed, 95% CI)	0.00 [‐0.46, 0.46]
1.7 Change in muscle strength (mean Medical Research Council sum score) within 6 weeks	2	42	Mean Difference (IV, Fixed, 95% CI)	0.78 [‐1.04, 2.60]
1.7.1 Parallel	1	28	Mean Difference (IV, Fixed, 95% CI)	0.29 [‐2.11, 2.69]
1.7.2 Cross‐over design	1	14	Mean Difference (IV, Fixed, 95% CI)	1.43 [‐1.35, 4.21]
1.8 Change in mean disability score at 24 weeks or later	1	117	Mean Difference (IV, Fixed, 95% CI)	0.80 [0.23, 1.37]
1.9 Frequency of serious side effects	3	315	Risk Ratio (M‐H, Fixed, 95% CI)	0.82 [0.36, 1.87]
1.9.1 Parallel design	2	258	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.30, 1.73]
1.9.2 Cross‐over design	1	57	Risk Ratio (M‐H, Fixed, 95% CI)	2.71 [0.12, 63.84]
1.10 Frequency of any side effects	3	308	Risk Ratio (M‐H, Fixed, 95% CI)	2.62 [1.81, 3.78]
1.10.1 Parallel design	2	258	Risk Ratio (M‐H, Fixed, 95% CI)	2.61 [1.80, 3.78]
1.10.2 Cross‐over design	1	50	Risk Ratio (M‐H, Fixed, 95% CI)	3.00 [0.13, 70.30]

Comparison 2. Intravenous immunoglobulin (IVIg) versus plasma exchange (PE).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Change in mean disability score within 6 weeks	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 3. Intravenous immunoglobulin (IVIg) versus prednisolone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Significant improvement in disability within 6 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.2 Improvement of ≥ 1 point on Rankin scale	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.3 Change in mean disability score within 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.4 Change in mean disability score on Rankin scale within 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.5 Change in muscle strength (mean Medical Research Council sum score) within 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.6 Frequency of serious side effects	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.7 Frequency of any side effects	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 4. Intravenous immunoglobulin (IVIg) versus intravenous methylprednisolone (IVMP).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Significant improvement in disability within 6 weeks	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
4.2 Improvement of ≥ 1 point on Rankin scale	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
4.3 Change in mean disability score within 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.4 Change in mean disability score on Rankin scale within 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.5 Change in muscle strength (Medical Research Council sum score) within 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.6 Change in mean disability score at 24 weeks or later	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
4.7 Frequency of serious side effects	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
4.8 Frequency of any side effects	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Camdessanche 2014.

Study characteristics
Methods	Design: randomised, open‐label study
Participants	CIDP criteria: unspecified Number of participants: 40 randomised, 5 excluded after randomisation, leaving 35 who received treatment (18 IVIg, 17 prednisone) Sex (M/F): 24/11, group assignment unspecified Age: mean 58 years Inclusion criteria Man or woman aged 18–80 years Weight ≤ 100 kg CIDP stable or deteriorated state (no spontaneous improvement), with the following features Motor or sensory and motor deficits Reduced or abolished tendon reflexes Progressive or relapsing evolution Global symmetric disability in > 1 limb Disease course installation over ≥ 2 months Cerebrospinal fluid with ≤10/µL white blood cells and > 0.5 g/L protein rate (non‐compulsory examination) Electrophysiological or histological signs of demyelination INCAT disability score ≥ 2 in arms or ≥ 1 in legs Exclusion criteria Severe electrophysiological axonal damage Pure motor syndrome Spontaneous improvement Associated systemic disease that could be the cause of neuropathy Severe cardiac insufficiency Cardiac arrhythmia Severe cardiopulmonary pathology Inflammatory syndrome Severe physical disease which can interfere with the trial Strict salt‐free diet Clinically significant abnormal biological result Positive serology in one of the following tests: HIV1, HIV2, A‐B‐C hepatitis, Hbs antigen Lyme disease IgA complete deficiency History of anaphylactic reaction during previous IVIg infusion Hypogammaglobulinemia (IgG < 3g/L) Creatinine clearance < 80 mL/min Evolutive gastroduodenal ulcer Diabetes Serious infectious condition Evolutive virus disease (hepatitis, herpes, varicella, zona) Psychotic states not controlled by treatment Veinous or arterial thrombosis Uncontrolled high blood pressure Osteoporosis Previous treatment with corticosteroids, IVIg, plasma exchanges or any other immunosuppressive agent within 3 months before inclusion, except for azathioprine and mycophenolate mofetil which were tolerated in the case of the dose being unmodified within 3 months and kept unchanged during the trial Previous failed treatment with IVIg or prednisone Hypersensitivity to any components of the 2 treatments Unsigned informed consent Ongoing or planned pregnancy (mandatory pregnancy test at the screening visit) Breastfeeding
Interventions	Experimental: IVIg (2g/kg), monthly for 6 months Control: prednisone (0.8 mg/kg), with a tapering regimen, for 6 months
Outcomes	Primary outcomes Response to treatment defined by ≥ 1 point less on the blinded INCAT score at 3 months in comparison to baseline Secondary outcomes Number of participants who recovered (INCAT upper limbs 0; lower limbs 1) at 3, 6, 9, and 12 months Tolerance of treatments
Sources of funding	LFB Biotechnologies
Conflicts of interest	Camdessanche has received research support from LFB Biotechnologies. Echaniz‐Laguna has received personal compensation for activities with Sanofi‐Aventis Pharmaceuticals, Inc. Stojkovic has received personal compensation for activities with LFB. All other authors had nothing to disclose.
Notes	Setting: multicentre study in 11 hospitals Location: France Date conducted: June 2004 to December 2013 Other: only abstract data available. Verified by communication with JP Camdessanche (11 November 2020).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information on random sequence generation available in abstract.
Allocation concealment (selection bias)	Unclear risk	No information on random sequence generation or allocation concealment available in abstract.
Blinding of participants and personnel (performance bias)	High risk	Open‐label study: participants and researchers both knew treatment assignment. Abstract mentions a blinded INCAT score for outcome assessment; no information provided in the abstract as to what this means and how this was accomplished.
Blinding of outcome assessment (detection bias)	High risk	Open‐label study: participants and researchers both knew treatment assignment. Abstract mentions a blinded INCAT score for outcome assessment; no information provided in the abstract as to what this means and how this was accomplished.
Incomplete outcome data (attrition bias) All outcomes	High risk	Abstract provides primary outcome data and a summary of results for secondary outcomes and severe adverse events. Five participants were excluded following randomization for unspecified reasons.
Selective reporting (reporting bias)	High risk	Trial is registered in clinicaltrials.gov but not protocol is available. Baseline characteristics were not provided. Primary outcome was reported without descriptive statistics. Secondary outcomes were not reported in full.
Other bias	Unclear risk	Data limited to the published abstract. No participant‐level data available following correspondence with trialists.

Dyck 1994.

Study characteristics
Methods	Design: single (observer)‐blind, cross‐over RCT
Participants	CIDP criteria: according to Dyck 1993 Number of participants: 20* (15 IVIg, 17 plasma exchange)* Sex (M/F): 10/9 (completing the 1st phase) Age: range 36–51 years; mean age (SD) at entry: plasma exchange 39 (17), IVIg 51 (18) Inclusion criteria Fulfilment of CIDP criteria Symptoms "severe enough" Weakness score ≥ 15 points (presumed to be on the NDS‐W) Neurological disability static or worsening No plasma exchange or IVIg in previous 6 weeks No other immunotherapy in previous 6 weeks Exclusion criteria Diseases known to predispose to neuropathy (alcoholism, myxoedema, diabetes mellitus, hepatic disease)
Interventions	Experimental: IVIg 0.4 g/kg/week for 3 weeks then 0.2 g/kg/week for another 3 weeks Control: plasma exchange 2/week for 3 weeks, then 1/week for 3 weeks
Outcomes	NDS Summated compound muscle action potential of median, ulnar and peroneal nerve (sumCMAP) Summated sensory nerve action potentials of median and sural nerve (sumSNAP) Vibration detection threshold of great toe
Sources of funding	These studies were supported in part by grants obtained from the National Institute of Neurological and Communicative Disorders and Stroke (14304) and from the Muscular Dystrophy Association. Immune globulin was provided by Miles Biological Products and Quantum Health Resources.
Conflicts of interest	None reported.
Notes	Setting: single‐centre study Location: Mayo clinic, Rochester, MN, USA Date conducted: not reported Other: predefined set of rules to determine length of wash‐out.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Inadequate information to judge sequence generation; study authors declined a request for additional information.
Allocation concealment (selection bias)	Unclear risk	Inadequate information to judge concealment; study authors declined a request for additional information.
Blinding of participants and personnel (performance bias)	Low risk	Participants and treatment providers were not blinded to treatment allocation, but this was unlikely to introduce bias.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "The neuropathic evaluations were generally performed by the same examiner, who was blinded as to the treatment used. Each patient was instructed by the patient coordinator not to reveal treatment information to the examiner. Additionally, the examiner performed the neurological examination before questioning the patient about symptoms." Comment: measures taken to assure blinding of the examiner seem to be adequate. Participants and treatment providers were not blinded to treatment allocation, but this was unlikely to introduce bias.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	1 participant withdrew during the first treatment period; neither treatment nor treatment effect were stated. No intention‐to‐treat analysis was performed. Seven participants did not cross over to the second treatment.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Low risk	No other bias identified.

Hahn 1996.

Study characteristics
Methods	Design: double‐blind, placebo‐controlled, cross‐over RCT
Participants	CIDP criteria: according to Ad hoc subcom 1991 Number of participants: 30* (25 IVIg; 25 placebo) Sex (M/F): 11/19 Age: mean 52 years (range 9–79 years) Inclusion criteria: Fulfilment of the diagnostic criteria of definite (all 4 criteria) or probable (3/4 criteria) CIDP (Ad hoc subcom 1991). For newly diagnosed patients, continuously progressive neuropathy for > 8 weeks For people with previously established diagnosis of chronic progressive or relapsing CIDP, clinical stasis or recent deterioration. Muscle weakness severe enough to interfere with secure ambulation (NDS = 40) Exclusion criteria Other diseases that could have produced the neuropathy, including monoclonal gammopathy of unknown significance (monoclonal paraproteins were determined by high resolution agarose gel serum and urine electrophoresis, Ig quantification and immunofixation techniques; localized myeloma was excluded by radiological skeletal survey) Previous IVIg treatment
Interventions	Experimental: IVIg 0.4 g/kg/day for 5 days Control: placebo (10% dextrose IV)
Outcomes	Modified NDS Clinical grading scale Grip strength Electrophysiological studies
Sources of funding	This study was supported by the Muscular Dystrophy Association of Canada, the Victoria Hospital Research Development Fund and the Miles/Canadian Red Cross Research and Development Fund.
Conflicts of interest	None reported.
Notes	Setting: Neuromuscular clinics Location: London, Ontario and Calgary, Alberta (Canada) Date conducted: prospective recruitment from 1990 to 1994 Other: Predefined set of rules to determine length of wash‐out. People who had been treated previously with various other forms of immunomodulatory therapy were permitted to take part in the trial.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Patients were randomly assigned to receive IVlg in standard dose or placebo infusions (10% dextrose) on 5 consecutive days." Comment: no details on randomisation procedure.
Allocation concealment (selection bias)	Unclear risk	Comment: no details on whether knowledge of the sequence was adequately concealed.
Blinding of participants and personnel (performance bias)	Low risk	Quote: "The participating patients, the evaluating neurologist, the electromyographer and the nurses administering care and infusions were blinded to the nature of the treatment during the controlled portion of the trial. ... To ensure that blinding was complete the infusions (Ivlg as well as placebo) were delivered from the transfusion services in identical opaque 600 ml transfer packs fitted with a sampling site coupler." Quote: "The code for individual patients was broken after completion of the second trial or at the time of trial analysis. Before the code was broken, study subjects and investigators were questioned regarding treatment effect and study phase assignments. However, a formal analysis of blinding was not performed." Comment: measures taken to assure blinding seem to be adequate.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "The participating patients, the evaluating neurologist, the electromyographer and the nurses administering care and infusions were blinded to the nature of the treatment during the controlled portion of the trial ... To ensure that blinding was complete the infusions (Ivlg as well as placebo) were delivered from the transfusion services in identical opaque 600 ml transfer packs fitted with a sampling site coupler." Quote: "The code for individual patients was broken after completion of the second trial or at the time of trial analysis. Before the code was broken, study subjects and investigators were questioned regarding treatment effect and study phase assignments. However, a formal analysis of blinding was not performed." Comment: measures taken to assure blinding seem to be adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Article provides sufficient details regarding 5 participants whose data were excluded from the analysis.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Unclear risk	Participants were permitted to take part in the trial despite other treatments if these treatments were not altered just before or during the trial. 3 participants did not cross over to the second treatment because of improvement during the first treatment period. All these participants were on the active treatment.

Hughes 2001.

Study characteristics
Methods	Design: double‐blind cross‐over RCT
Participants	CIDP criteria: well‐defined criteria, including INCAT Neurophysiological Criteria for Diagnosis of CIDP Number of participants: 32* (24 IVIg; 24 prednisolone) Sex (M/F): 21/11 Age: mean 54.1 (SD 17.1) years Inclusion criteria Age ≥ 18 years Diagnosis of CIDP by a consultant neurologist: Progressive or relapsing motor and sensory dysfunction of more than one limb resulting from neuropathy developing over 2 months Reduced or absent tendon reflexes CSF < 10 white cells/mL (CSF examination encouraged but not mandatory) Fulfilment of INCAT neurophysiological criteria for multifocal demyelinating polyradiculoneuropathy Significant disability in upper or lower limb function (at least arm disability grade 2 or leg disability grade 1) Stable or worsening clinical condition (not improving spontaneously) Exclusion criteria Associated systemic diseases that might cause neuropathy, such as carcinoma, systemic lupus erythematosus, diabetes mellitus, paraproteinemia (detectable by conventional high‐resolution agarose gel electrophoresis), or HIV infection Pregnancy or planned pregnancy Severe concurrent medical conditions or previous neurological deficit that would interfere with treatment, including active peptic ulceration, renal failure, known requirement for steroids, or known allergic reaction to IVIg Significant respiratory impairment Treatment with steroids, IVIg, plasma exchange, or any immunosuppressant drug during the 6 weeks before randomisation (treatment with azathioprine permitted provided that the dose has been unchanged for 6 months and expected to remain unchanged) Pure motor syndrome fulfilling criteria for multifocal motor neuropathy with conduction block Previous failure to respond to either IVIg or prednisolone
Interventions	Experimental: IVIg 1.0 g/kg/day for 2 days or 2.0 g/kg over 24 hours Control: prednisolone 60 mg for 2 weeks, 40 mg for 1 week, 30 mg for 1 week, 20 mg for 1 week, 10 mg for 1 week
Outcomes	INCAT disability scale Timed 10‐metre walk 9‐hole pegboard test MRC sum score Maximum grip strength Rotterdam Handicap scale Rankin scale SF‐36 quality of life scale Electrophysiological studies
Sources of funding	Novartis provided the Sandoglobulin and placebo infusions. Novartis and the Guillain‐Barré Syndrome Support Group provided financial support.
Conflicts of interest	None reported.
Notes	Setting: multicentre (9 European sites) Location: Europe Date conducted: July 1998–November 1999 Other: predefined set of rules to determine length of washout
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomly allocated to the treatment sequences, prednisolone followed by IVIg or IVIg followed by prednisolone. The sequences were stratified by center with a block size of 2 according to a sequence of random numbers provided by the trial statistician to the manufacturers who prepared the randomization packs."
Allocation concealment (selection bias)	Low risk	Quote: "The block size was not revealed to the trialists. Each center was provided with coded packs containing either Sandoglobulin or placebo and tablets containing either placebo or prednisolone. Novartis prepared the coded packs from a random sequence generated by the trial statistician that was not revealed to the investigators except in an emergency." Comment: measures taken to assure concealment of allocation seem to be adequate.
Blinding of participants and personnel (performance bias)	Low risk	Quote: "Placebo infusions contained albumin 6 mg/100 ml and were identical in appearance to the IVIg infusions. A neurologist who did not have access to laboratory data or the liberty to ask questions about adverse events assessed patients at entry and after 2, 4, and 6 weeks into each treatment period. A different neurologist applied a questionnaire about adverse events after 2, 4, and 6 weeks." Comment: measures taken to assure blinding of participants and personnel seem to be adequate.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "A neurologist who did not have access to laboratory data or the liberty to ask questions about adverse events assessed patients at entry and after 2, 4, and 6 weeks into each treatment period. A different neurologist applied a questionnaire about adverse events after 2, 4, and 6 weeks." Comment: measures taken to assure blinding of outcome assessment seemed to be adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Results from all participants whose data were not used for primary analysis have been addressed (primary analysis was performed only for participants completing both treatments). Not all participants entered the cross‐over phase.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Low risk	No other bias identified.

Hughes 2008.

Study characteristics
Methods	Design: double‐blind, response‐conditional, cross‐over RCT with extension phase
Participants	CIDP criteria: INCAT Number of participants: 117 (59 IVIg, 58 placebo) Sex: M/F: total: 77/40; IVIg group: 31/28; placebo group: 46/12 Age: range 18–83 years; IVIg group: mean 50 (SD 17) years; placebo group: 53 (SD 16) years Inclusion criteria Diagnosis of CIDP with progressive or relapsing motor and sensory dysfunction of ≥ 1 limb resulting from neuropathy over the 2 months before study entry Significant disability as defined by an overall INCAT disability score of 2–9 (INCAT disability score of 2 had to be exclusively from leg disability to be eligible) Exclusion criteria Treatment with steroids (> 10 mg/day prednisolone or equivalent), IVIg, or plasma exchange in the 3 months before study entry Use of fish‐oil supplements in the previous month Treatment with other immunomodulatory or immunosuppressive agents (interferon or azathioprine) in the previous 6 months Myelopathy or evidence of central demyelination Persistent neurological deficits from stroke, CNS trauma, or peripheral neuropathy from other causes (e.g. diabetes mellitus; IgM paraproteinaemia; or uraemic, toxic, or familial neuropathy) A motor syndrome that fulfilled criteria for MMN with conduction block (i.e. upper limb motor weakness without sensory deficit and with a 50% decrease in action potential amplitude or area on proximal compared with distal stimulation in motor nerves) Evidence of systemic disease that might cause neuropathy
Interventions	Experimental: IVIg 2.0 g/kg over 2 to 4 days and maintenance 1.0 g/kg over 1 to 2 days every 3 weeks for up to 24 weeks Control: placebo
Outcomes	INCAT disability scale Maximum grip strength CMAP amplitude of the most severely affected motor nerve Time to relapse
Sources of funding	Talecris Biotherapeutics Center for Science and Education funded the study.
Conflicts of interest	2 authors worked for Talecris Biotherapeutics. 7 authors received honoraria for participation on the ICE Study Steering Committee. 1 author received hospitality and his department received consultancy fees from Talecris Biotherapeutics. 1 author received honoraria from Talecris Biotherapeutics for speaking at scientific symposia and for serving on a steering committee.
Notes	Setting: multicentre study Location: 33 centres in Europe, North America, South America, and Israel Date conducted: recruitment April 2004–June 2005 Other: first trial period had a parallel‐group design with a conditional cross‐over 'escape' for participants who did not improve sufficiently.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Computer‐generated random codes and treatment assignments were prepared by an independent group within the sponsor hierarchy and were distributed by the sponsor to the unblinded pharmacist at each centre." Comment: we consider measures taken to assure adequate sequence generation to be adequate.
Allocation concealment (selection bias)	Low risk	Quote: "Eight randomisation numbers, in four blocks of two random numbers each, were initially assigned to each centre (block size was not disclosed to the centres). If a centre required additional random numbers, it received a set of eight numbers. The same procedure was used to generate separate random codes to assign patients to placebo or IGIV‐C in the extension phase." Comment: we consider measures taken to assure concealment of allocation to be adequate.
Blinding of participants and personnel (performance bias)	Low risk	Quote: "During the study, unblinded monitors checked the drug batch log to ensure that the study medication was prepared and given as assigned. All other study team members were blinded to patient treatment during the study." Comment: we consider measures taken to assure blinding to be adequate.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "During the study, unblinded monitors checked the drug batch log to ensure that the study medication was prepared and given as assigned. All other study team members were blinded to patient treatment during the study." Comment: we consider measures taken to assure blinding to be adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Low risk	The study appears to be free of other sources of bias.

Mendell 2001.

Study characteristics
Methods	Design: double‐blind, parallel‐group RCT
Participants	CIDP criteria: Ad hoc subcom 1991 Number of participants: 50 (29 IVIg; 21 placebo) Sex (M/F): IVIg group: 16/14; placebo group 11/12 Age: IVIg group: mean 54 (SD 20) years, range 13–82 years; placebo group: mean 50 (SD 18) years, range 23–73 years Inclusion criteria Fulfilment of diagnostic criteria for definite or probable CIDP (Ad Hoc Subcommittee of the American Academy of Neurology) Exclusion criteria Concomitant disease (myelopathy or evidence of central demyelination, pre‐existing endocrinopathy, connective tissue disease, HIV infection, hepatitis, Lyme disease, cancer, paraproteinemia, or Castleman's disease) Pregnancy or breastfeeding Elevated serum creatinine Treatment for any disease requiring immunomodulatory agents (IVIg, cyclophosphamide, azathioprine, corticosteroids, tacrolimus, cyclosporine, OKT3, plasma exchange, alpha, beta, or gamma interferon) within the past 3 months
Interventions	Experimental: IVIg 1.0 g/kg/day for 2 days Control: placebo (5% albumin IV)
Outcomes	Hughes functional disability scale Average muscle score Forced vital capacity Electrophysiological studies
Sources of funding	Supported by NIH‐sponsored General Clinical Research Center (GCRC) at Ohio State University (grant no. MO1‐RR00034). Also supported by the GCRC units at University of Texas Southwestern (grant no. MO1‐RR00633) and the Audie L. Murphy Veterans Hospital (gran no. MO1‐RR01246). The IV immunoglobulin (Pasteurized Gammar IV) was supplied by Aventis Behring, LLC, King of Prussia, PA.
Conflicts of interest	None reported.
Notes	Setting: multicentre study organised by the Working Group on Peripheral Neuropathy with IVIg provided by LLC, King of Prussia, PA. Location: North America Date conducted: not mentioned; conducted over a predesignated 3‐year period. Other: none
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A computer‐generated randomization plan stratified subjects in blocks of four at each site' Comment: measures taken to assure random sequence generation seem to be adequate.
Allocation concealment (selection bias)	Low risk	Quote: "A computer‐generated randomization plan stratified subjects in blocks of four at each site' Comment: random sequence generation seems to be adequate; it seems unlikely that assignment of participants could be predicted.
Blinding of participants and personnel (performance bias)	Low risk	Quote: "To preserve the blind, all study medications were prepared by the pharmacy at each center and delivered to the infusion unit enclosed in amber IV bags." Comment: we consider measures taken to assure blinding to be adequate.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "To preserve the blind, all study medications were prepared by the pharmacy at each center and delivered to the infusion unit enclosed in amber IV bags." Comment: we consider measures taken to assure blinding to be adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "In addition, no postinfusion efficacy data were available for two patients in the placebo group: one developed urticaria during the first infusion and the other dropped out after the first infusion because of personal preference. One patient treated with IVIg was dropped from the analysis before the unblinding because of an error in record documentation at the site." Comment: no intention‐to‐treat analysis possible due to withdrawal of 3 participants. Reasons for withdrawal were unlikely to be associated with disease or treatment.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Unclear risk	There was an imbalance in recruitment and a prestudy design for block randomisation by participating centres, which accounted for the discrepancy in the number of participants in the groups.

Nobile‐Orazio 2012.

Study characteristics
Methods	Design: double‐blind, placebo‐controlled, parallel‐group, multicentre RCT
Participants	CIDP criteria: INCAT Number of participants: 46 (25 IVIg, 21 IVMP) Sex (M/F): methylprednisolone: 15/6; IVIg: 15/9 Age: IVMP: mean 66 years (range 39–79); IVIg: mean 54 years (range 19–83) Inclusion criteria Age ≥ 18 years Definite typical CIDP according to the EFNS/PNS criteria with disability (≥ 2 on either the ONLS or the modified Rankin scale Active or stationary phase but not in remission compared with the last available assessment No improvement in the ONLS and modified Rankin scale scores between the screening and inclusion visits Exclusion criteria Atypical CIDP A diagnosis of multifocal motor neuropathy other underlying causes including diabetes and IgM monoclonal gammopathy with antibodies to myelin associated glycoprotein or to sulfatides Concurrent medical disorders preventing treatment or assessment History of psychiatric disorders Abnormalities on chest radiography Ocular hypertension Known contraindications to steroids or to IVIg Use of IVIg in the previous 8 weeks or corticosteroids in the previous 3 months at a dose > 25 mg on alternate days or 12.5 mg daily (lower doses were accepted but must not have changed in the previous 3 months) Documented lack of response to a previous course of an effective dose of steroids (methylprednisolone 0.5 g IV for 3 days or oral corticosteroids at a dose comparable to 1 mg/kg per day of prednisone for at least 1 month) or to IVIg 2 g/kg (concurrent immune therapy was accepted if not changed in the previous 12 months) Ongoing or planned pregnancy Increased concentrations of cells in their CSF (> 10/mL)
Interventions	Experimental: IVIg (0.5 g/kg/day) over 4 days, every month for 6 months Control: IVMP 0.5 g/kg/day over 4 consecutive days, every month for 6 months
Outcomes	Proportion of participants discontinuing treatment due to inefficacy, intolerance, or adverse events during a 6‐month follow‐up ONLS disability scale Rankin score MRC sum score Grip strength INCAT sensory sum score Vibratory score Time on 10‐metre walk Rotterdam score SF‐36 quality of life
Sources of funding	Kedrion
Conflicts of interest	1 author served on the Immunoglobulin in Neurology Advisory Board of CSL Behring and as ad‐hoc consultant for Baxter, Laboratoire Français du Biofractionnement, and Kedrion, and lectured for Talecris. 2 authors received travel grants to attend scientific meetings from Baxter, Grifols, Kedrion, and Novartis. 2 authors received travel grants to attend scientific meetings from Grifols and Kedrion. 6 authors received travel grants to attend scientific meetings from Kedrion. 1 author received travel grants to attend scientific meetings from Baxter, CSL Behring, and Kedrion. 1 author received travel grants to attend scientific meetings from Baxter, Grifols, and Kedrion. 1 author received travel grants to attend scientific meetings from Biofutura, Grifols, Kedrion, and Lilly. 2 authors received travel grants to attend scientific meetings from Grifols. 1 author served on the advisory board of Viropharma and lectured for UCB‐Pharma, Sanofi ‐Aventis, Eisai, and GSK. 1 author's institution received grants from Sanofi ‐Aventis, Janssen, and Eisai. 1 author was an employee of Kedrion. 1 author declared no conflicts of interest.
Notes	Setting: multicentre, enroled from 14 Italian neurological centres Location: Italy Date conducted: September 2007–December 2009 Other: participants with lack of improvement within 2 months were switched to the alternative treatment.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Allocation to treatment was stratified by centre and centrally managed with a computer‐generated 1:1 randomisation scheme with a sequential block size of four" Comment: random sequence generation seems adequate.
Allocation concealment (selection bias)	Low risk	Quote: "Allocation to treatment was stratified by centre and centrally managed with a computer‐generated 1:1 randomisation scheme with a sequential block size of four" Comment: random sequence generation seems to be adequate, it seems unlikely that assignment of participants can be predicted.
Blinding of participants and personnel (performance bias)	Low risk	Quote: "Patients, their families, and investigators were masked to treatment assignment." Comment: many participants had previously used the trial treatments, leading to possible unblinding due to previously encountered adverse events.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "Patients, their families, and investigators were masked to treatment assignment." Comment: many participants had previously used the trial treatments, leading to possible unblinding due to previously encountered adverse events.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data were available for all included participants.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Low risk	No other bias identified.

Thompson 1996.

Study characteristics
Methods	Design: double‐blind cross‐over RCT
Participants	CIDP criteria: Ad hoc subcom 1991 Number of participants: 7* (7 IVIg; 7 placebo) Sex (M/F): 6/1 Age: mean 46 years (range 26–65 years) Inclusion criteria Meeting the clinical, neurophysiological and CSF criteria for CIDP Exclusion criteria Associated systemic diseases A serum paraprotein
Interventions	Experimental: IVIg 0.4 g/kg/day for 5 days Control: placebo (albumin solution)
Outcomes	Hammersmith Motor Ability Score Ambulation index 10‐mentre walk time Expanded MRC sum score Myometry 9‐hole Peg Test
Sources of funding	The Muscular Dystrophy Group of Great Britain, the Guillain‐Barré Syndrome Support Group of the United Kingdom, and the Wellcome Research Foundation provided financial support.
Conflicts of interest	None reported.
Notes	Setting: single‐centre study Location: UK Date conducted: not mentioned Other: predefined set of rules to determine length of washout
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Patients were randomly treated with infusions of either IVIg" Comment: No details of randomisation procedure.
Allocation concealment (selection bias)	Unclear risk	Comment: no details of allocation concealment.
Blinding of participants and personnel (performance bias)	Low risk	Quote: ''Patients were randomly treated with infusions of either IVIg ... under double masked conditions". Comment: although no details on blinding are reported, blinding seems to be adequate.
Blinding of outcome assessment (detection bias)	Low risk	Quote: ''Patients were randomly treated with infusions of either IVIg ... under double masked conditions". Comment: although no details on blinding are reported, blinding seems to be adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data were available for all included participants.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Unclear risk	No mention of any additional treatments. Primary outcome assessed at two weeks; we consider this is not a clinically appropriate study duration.

Vermeulen 1993.

Study characteristics
Methods	Design: double‐blind, parallel‐group, placebo‐controlled, multicentre RCT
Participants	CIDP criteria: Ad hoc subcom 1991 Number of participants: 28 (15 IVIg; 13 placebo) Sex (M/F): IVIg: 4/11; placebo: 4/9 Age IVIg group: mean 45 years: placebo group: mean 50 years Inclusion criteria Symptoms and signs of polyneuropathy in the absence of systemic disease Electrophysiological diagnosis of demyelinating polyneuropathy (based on slowed nerve conduction velocities or conduction blocks, increased CSF protein (> 05 g/L), and progression of weakness exceeding 8 weeks Normal ESR, haematocrit, white cell and platelet count, serum creatinine, serum glucose, normal liver and thyroid function tests, no antinuclear antibodies, cryoglobulin, or monoclonal protein and also a normal chest radiograph Disability of ≥ 3 on the modified Rankin scale Exclusion criteria Kinship history of neuropathy Immunosuppressive treatment
Interventions	Experimental: IVIg 0.4 g/kg/day for 5 days Control: placebo (60 mg/ml albumin solution)
Outcomes	Rankin scale MRC sum score Electrophysiological studies
Sources of funding	This work was supported by grants from the Prinses Beatrix Fonds and Stichting Willem H Kroger.
Conflicts of interest	None reported
Notes	Setting: multicentre study Location: the Netherlands Date conducted: not mentioned Other: none
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "When a patient was eligible and after informed consent, the Central Laboratory of the Netherlands Red Cross Blood Transfusion Service Amsterdam (CLB) was informed. The CLB supplied either bottles with immunoglobulin or placebo for a complete treatment course, according to a list based on a random number table, to the centre where the patient was admitted." Comment: randomisation method appears adequate.
Allocation concealment (selection bias)	Low risk	Quote: "The CLB supplied either bottles with immunoglobulin or placebo for a complete treatment course, according to a list based on a random number table, to the centre where the patient was admitted. " Comment: measures taken to assure concealment of allocation seem to be adequate; it seems unlikely that future assignment of participants could be deduced from the random number table (kept by the CLB).
Blinding of participants and personnel (performance bias)	Low risk	Quote: "Contents, size and labels of the bottles were not distinguishable. The trial code was broken after the results of all patients had been recorded." Comment: measures taken to assure blinding seem to be adequate.
Blinding of outcome assessment (detection bias)	Low risk	Quote: "Contents, size and labels of the bottles were not distinguishable. The trial code was broken after the results of all patients had been recorded." Comment: measures taken to assure blinding seem to be adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcomes stated in methods section were reported for all participants.
Selective reporting (reporting bias)	Low risk	All outcomes stated in methods section were reported in results section.
Other bias	Unclear risk	The primary outcome was assessed between day 16 and 21, just below the cut‐off level of 3 weeks, which we considered the lower limit of a clinically appropriate study duration (and outcome assessment). Quote: "Patients eligible for this study were admitted with symptoms and signs of polyneuropathy in the absence of systemic disease, with an electrophysiological diagnosis of demyelinating polyneuropathy based on slowed nerve conduction velocities and or conduction blocks, increased CSF protein (more than 0.5 g/l) and progression of weakness exceeding eight weeks." Comment: no electrophysiological criteria were provided for the definition of a demyelinating polyneuropathy.

CIDP: chronic inflammatory demyelinating polyradiculopathy; CNS: central nervous system; CSF: cerebrospinal fluid; EFNS/PNS: European Federation of Neurological Societies/Peripheral Nerve Society; ESR: erythrocyte sedimentation rate; INCAT: Inflammatory Neuropathy Cause and Treatment; IV: intravenous; IVIg: intravenous immunoglobulin; IVMP: intravenous methylprednisolone; MMN: multifocal motor neuropathy; MRC: Medical Research Council; NDS: Neuropathy Disability Score; NDS‐W: Neuropathy Disability Score‐weakness subset; ONLS: overall neuropathy limitation scale; RCT: randomised controlled trial; SD: standard deviation; SF‐36: 36‐item Short Form Health Survey; wk: week.

*In cross‐over trials, the number of randomised participants does not equal the number of participants in the treatment arms. For an explanation of these discrepancies, we refer the reader to the Results section of this review.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Baba 1996	No randomisation procedure. Diagnostic criteria and outcome measures not mentioned.
Breiner 2019	Ineligible patient population. All participants had diabetes mellitus, a participant exclusion criterion.
Cornblath 2022	Ineligible comparator (IVIg). Compared low and high dose maintenance IVIg treatment to regular (1 g/kg) IVIg maintenance treatment in previously treated, proven IVIg‐dependent people.
Curro 1987	No randomisation procedure.
Dalakas 1996	No definite or probable CIDP patients included.
EudraCT2012‐001996‐34	Ineligible comparator (IVIg). Compared 2 different IVIg preparations in previously (IVIg) treated participants.
Hankey 1994	'n of 1' trial. Outcome measures did not include a disability scale.
ISRCTN13637698	Ineligible comparator. Selection bias: patient population consisted of IVIg‐treated patients.
Kubori 1999	Not properly randomised: lacking a placebo/control group. Participants were divided depending on clinical severity in 3 groups and treated with different dosages of IVIg. CIDP and MMN patients were not separated in the analysis.
Kuitwaard 2021	Ineligible comparator. Selection bias: population consisted of IVIg‐treated patients.
Kuwabara 2017	Non‐randomised; open‐label design.
Léger 2013	Non‐randomised; open‐label design.
Markvardsen 2017	Ineligible intervention (SCIg). Trial potentially eligible for SCIg Cochrane Review.
NCT01225276	Study terminated.
NCT03684018	Ineligible comparator.
Neacsu 2016	Unavailable for full‐text review. Appeared to have non‐randomised design.
van Doorn 1990b	Based on participant population, only included known IVIg responders.
Zinman 2005	Excluded based on comparator.

CIDP: chronic inflammatory demyelinating polyradiculopathy; IVIg: intravenous immunoglobulin; MMN: multifocal motor neuropathy; SGIg: subcutaneous immunoglobulin.

Differences between protocol and review

See Van Schaik 1999 (protocol).

2013 update

The search of ISI did not add value to other searches, so we stopped searching the ISI database.

We added risk of bias tables (Higgins 2011), and summary of findings tables (Eftimov 2013).

2023 update

We separated blinding into two domains: Blinding of participants and personnel (performance bias) and Blinding of outcome assessment (detection bias), as per the 2017 Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We also documented that we resolved differences in risk of bias assessments by consensus.

John Winer retired from the review author team and did not participate in the 2023 update.

We included an additional table summarising the characteristics of all included studies (Table 6).

To improve clarity, we reduced the number of outcomes reported in the summary of findings table.

We documented use of the Rankin scale as alternative reporting for the primary outcome, without changing the approach from earlier versions of the review.

We excluded one study that was previously included because the comparator was immunoadsorption (not listed as an eligible comparator for this review).

We excluded studies that enroled participants based on prior intravenous immunoglobulin (IVIg) response. We revised the Methods section to reflect this change.

We changed our choice of model to the more conservative random‐effects analysis, which is now preferred by Cochrane Neuromuscular. We performed a sensitivity analysis using a fixed‐effect model in the presence of heterogeneity.

We edited the body of the review to reflect updated reporting guidelines (Higgins 2022c). Specifically, in the Methods section, we:

• reported that we would arrange for translation of studies where required;

• stated that studies including a subset of participants that matched our inclusion criteria were eligible for the review if separate data for the eligible subset were available;

• documented that we asked study authors for original data;

• reported that we did not use outcomes as inclusion criteria, and we included studies in the review irrespective of whether measured outcome data were reported in a 'usable' way;

• documented that we would include only eligible intervention arms from multiarm studies and explained how we would avoid double‐counting of participants in such studies;

• documented that we used standardised men difference (SMD) as the effect measures for comparability with other analyses where appropriate (this was also the approach in previous versions of the review);

• included the I² test in assessment of heterogeneity and specified a significance level of P > 0.1 for tests for heterogeneity;

• stated that we would undertake meta‐analysis only when treatments, participants, and the clinical question were similar enough for pooling to make sense; and

• added more detail on the GRADE assessment process.

Contributions of authors

Previous versions: INvS, JBW (JB Winer), MV, RdH were involved in all aspects of the development of the protocol. INvS and JBW assessed all the trials, extracted data, and discussed and resolved any problems for the original version of the review. FE, JBW, and INvS performed these tasks for the 2009 and 2013 updates. FE, RdH, and INvS calculated the effect sizes and did the meta‐analysis. INvS wrote the first version of the review, which all other review authors evaluated and amended.

FE wrote the 2013 update, which all other review authors evaluated and amended.

SB wrote the 2023 update, which all other review authors evaluated and amended.

Sources of support

Internal sources

• None, Other

  No internal sources of support for this update.

External sources

• National Institutes of Health Research, UK

  This project was supported by the National Institute for Health Research via Cochrane Infrastructure funding to Cochrane Neuromuscular. The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, NHS or the Department of Health.

  Cochrane Neuromuscular is also supported by the Queen Square Centre for Neuromuscular Disease.

Declarations of interest

SB: reports no competing interests.
RdH: reports no competing interests.
MV: reports no competing interests.
INvS: chairs a steering committee for CSL‐Behring and received departmental honoraria for serving on scientific advisory boards for CSL‐Behring and UCB. He received speakers' fees from CSL‐Behring. Departmental research support has been granted by The Netherlands Organisation for Scientific Research, and the Dutch Prinses Beatrix Spierfonds. All lecturing and consulting fees for INvS were donated to the Stichting Klinische Neurologie, a local foundation that supports research in the field of neurological disorders. INvS is a member of the Scientific Board of the Kreuth III meeting on the optimal use of plasma‐derived medicinal products, especially coagulation factors and normal immunoglobulins organised under the auspices of the European Directorate for the Quality of Medicines & HealthCare (EDQM).
FE: received support for printing of his thesis from Sanquin Bloedvoorziening (Dutch blood bank and IVIg manufacturer) and CSL‐Behring in 2014. He reports lecture fees from CSL‐Behring, Kedrion, and Grifols. Outside the submitted work, as principal investigator of INCbase, FE also reports investigator‐initiated grants from Kedrion, Terumo BCT, CSL‐Behring, and Takeda Pharmaceutical Company, and grants from ZonMw (Dutch governmental agency) and Prinses Beatrix Spierfonds (a Dutch charity). In addition, his institution has received fees from UCB Pharma, CSL‐Behring, and Takeda for advisory board membership. All grants and fees were paid to his institution. He is a member of the Cochrane Neuromuscular Editorial Board.

Edited (no change to conclusions)