Amniotic membrane transplantation for acute ocular burns

Abstract

Background

Ocular surface burns can be caused by chemicals (alkalis and acids) or direct heat. One effect of the burn is damage to the limbal epithelial stem cells of the ocular surface with delayed re‐epithelialisation, stem cell failure, and conjunctivalisation of the cornea. Amniotic membrane transplantation (AMT) performed in the acute phase (day 0 to day 7) following an ocular surface burn is claimed to reduce pain and accelerate healing. The surgery involves securing a layer of amniotic membrane (AM) to the eyelid margins as a patch to cover the entire ocular surface. However, there is debate about the severity of an ocular burn that may benefit from AMT and uncertainty of whether AMT improves outcomes.

Objectives

To compare the effect of AMT with medical therapy in the first seven days after an ocular surface burn, compared to medical therapy alone.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL; which contains the Cochrane Eyes and Vision Trials Register; 2021, Issue 9); Ovid MEDLINE; Ovid Embase; LILACS; the ISRCTN registry; ClinicalTrials.gov and the WHO ICTRP. We did not use any date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 29 September 2021.

Selection criteria

We included randomised trials that compared an AMT applied in the first seven days following an ocular surface burn in addition to medical therapy with medical therapy alone. The outcome measures were failure of re‐epithelialisation by day 21 post injury, visual acuity at final follow‐up, corneal neovascularisation, symblepharon, time to re‐epithelialisation and adverse effects.

Data collection and analysis

Two review authors independently screened search results, assessed the included studies for risk of bias and extracted relevant data. We contacted trial investigators for missing information. We summarised data using risk ratios (RRs) and mean differences (MDs) as appropriate.

Main results

We analysed two RCTs, but excluded individual patients who had been treated outside the acute phase in one of the studies (data provided by study authors). In total, 36 moderate burns from one RCT and 92 severe burns from two RCTs were evaluated separately. For both categories, the certainty of the evidence was downgraded principally as a result of high risks of performance and detection biases, and because of imprecision indicated by very wide confidence intervals. In addition, follow‐up was insufficiently frequent to calculate time‐to‐epithelialisation precisely.

Moderate severity ocular burns (Roper‐Hall classification II‐III)

The relative risk of AMT on failure of epithelialisation by day 21 was 0.18 (0.02 to 1.31), and LogMAR visual acuity was 0.32 lower (0.55 to 0.09 lower) in the treatment group (i.e. better), suggesting a possible benefit of AMT. The GRADE assessment for failure of epithelialisation by day 21 was downgraded to very low due to the risk of bias and imprecision (very wide confidence intervals including no effect). The GRADE assessment for visual acuity at final follow‐up was downgraded to low due to the risk of bias and imprecision (optimal information size not met). The relative effects of AMT on corneal neovascularisation (RR 0.56; 0.21 to 1.48), symblepharon (RR 0.41; 0.02 to 9.48) and time‐to‐epithelialisation (13 days lower; 26.30 lower to 0.30 higher) suggest possible benefit of AMT, but the wide confidence intervals indicate that both harm and benefit are possible. GRADE assessments for these outcomes were once again downgraded to very low due to the risk of bias and imprecision. Since adverse effects are rare, the small sample would have fewer occurrences of rare but potentially important adverse effects. The GRADE assessment for adverse effects was therefore considered to be low. 

Severe ocular burns (Roper‐Hall classification IV)

The relative risk of AMT on failure of epithelialisation by day 21 was 1.03 (0.94 to 1.12), and LogMAR visual acuity was 0.01 higher (0.29 lower to 0.31 higher) in the treatment group (i.e, worse), indicating no benefit of AMT. GRADE assessments for failure of epithelialisation by day 21 and final outcomes were downgraded to low. The relative effects of AMT on corneal neovascularisation (RR 0.84; 0.66 to 1.06), symblepharon (RR 0.89; 0.56 to 1.42) and time‐to‐epithelialisation (1.66 days lower; 11.09 lower to 7.77 higher) may include both benefit and harm. GRADE assessments for corneal neovascularisation, symblepharon and time‐to‐epithelialisation were downgraded to low due to risk of bias and imprecision. For adverse effects, the GRADE assessment was downgraded to low, reflecting the small sample sizes in the RCTs.

Authors' conclusions

There is uncertain evidence to support the treatment of moderate acute ocular surface burns with AMT in addition to standard medical therapy as a means of preventing failure of epithelialisation by day 21, improving visual outcome and reducing corneal neovascularisation, symblepharon formation and time‐to‐epithelialisation. For severe burns, the available evidence does not indicate any significant benefit of treatment with AMT.

Plain language summary

Amniotic membrane transplantation for the treatment of ocular burns

How an ocular burn damages the eye

The cornea is the transparent window of the eye that helps focus light to form an image of the outside world. For the cornea to remain transparent, it must have a healthy surface layer of epithelium, a stable tear film, and an absence of blood vessels. The epithelium is the moist cellular layer that covers the cornea, the white part of the eye, and the inside of the eyelids. The epithelium that covers the cornea is different from the epithelium of the conjunctiva. It is maintained in this state by a population of specialised stem cells found at the margin of the cornea (the limbus). In the normal eye, corneal epithelial cells slowly migrate from the limbus towards the centre of the cornea.

A burn to the eye initially damages the surface epithelial layer, but it can also damage the deeper structures of the eye and the eyelids. A characteristic of an acute burn is an epithelial defect. This defect can fully recover after a minor burn. However, if the burn damages the stem cells of a segment of the limbus, the normal epithelium central to that area may not recover. The cornea is then covered by a functionally different type of epithelium derived from the peripheral conjunctiva, which may take weeks or months, and the cornea can become hazy and vascularised. In the absence of any epithelial covering, an additional risk is that the deeper tissues of the exposed cornea (stroma) may become progressively thinner and eventually perforate. The more extensive the damage to the limbus (measured in clock hours), the worse the outcome. The abnormal surface of the cornea, as well as haze and damage to structures within the eye (iris, lens, retina, optic nerve), can all lead to loss of vision.

Treatments for ocular burns

A burn to the surface of the eye is an emergency, and the eye must be washed immediately with water or saline to remove any chemical agents. The amount of damage to the epithelium (an epithelial defect) and the blood vessels at the edge of the cornea are measured, estimating the risk of future complications. The eye is then treated with medicines to reduce the risk of further complications and help the epithelial layer recover. Some doctors also use an amniotic tissue membrane placed over the surface of the eye. The amniotic membrane is the sac that surrounds a baby as it develops inside its mother.

Why we did this Cochrane Review

We wanted to know whether using an amniotic membrane with medicines is better than just medicines.

What did we do?

We looked for studies in which an amniotic membrane was used to treat eyes that had been burned. We then looked for studies comparing eyes treated with medicines and an amniotic membrane patch with eyes treated using only medicines. We looked for randomised controlled trials, in which the treatment that people received was decided at random because these studies give the most reliable evidence about the effect of treatment. We assessed the treatment of moderate burns and severe burns separately, because they have different outcomes. We measured whether the treatment worked by comparing the proportion of eyes in which the epithelium over the cornea had failed to recover 21 days after the injury, visual function at the end of treatment, the presence of abnormal new blood vessels on the cornea, scarring between the cornea and the lids, the time taken for the epithelium to heal and adverse events in the intervention and control groups. We assessed the risk of bias in the randomised controlled study findings across pre‐specified domains. We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) criteria to evaluate certainty of evidence include risk of bias, imprecision, inconsistency, indirectness, and publication bias. 

What we found

We found two randomised controlled studies comparing the effect of medicines with the addition of an amniotic membrane patch inserted within seven days of the injury against medicines alone. One study suggested that treatment with amnion may be beneficial in moderate burns, but we found the certainty of the evidence to be low due to a high risk of bias and serious imprecision. For severe burns, there were no important differences between eyes that received an amniotic membrane and eyes that did not. 

Conclusions

There is only uncertain evidence from randomised controlled studies of an important difference in healing between eyes treated with medicines and an amniotic membrane within seven days of an ocular burn and eyes that are treated only with medicines. Further research is needed to identify the best treatment for acute burns to the eye to reduce the risk of potentially devastating consequences and visual loss.

How up to date is this review?
Searches were last done in September 2021.

Summary of findings

Summary of findings 1. Amniotic membrane transplant (AMT) for moderate ocular burns.

Amniotic membrane transplant (AMT) for moderate ocular burns
Patient or population: people with moderate acute ocular burns Settings: ophthalmic hospital Intervention: AMT and medical therapy Control: medical therapy alone
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of Participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	AMT
Failure of re‐epithelialisation by day 21 post‐injury  Image analysis of digital photographs Follow‐up: 6 months to 24 months	350 per 1000	63 per 1000  (7 to 459)	RR 0.18  (0.02 to 1.31)	36 eyes (1 RCT)	⊕⊝⊝⊝ very low1,2
**Visual acuity at final follow‐up  LogMAR Scale from 0 to 3 (Lower values = better vision)  Follow‐up: 6 months to 24 months	The mean VA at final follow‐up in the control group was 0.38	The mean VA at final follow‐up in the intervention group was 0.32 lower  (0.55 to 0.09 lower)		36 eyes (1 RCT)	⊕⊕⊝⊝ low1,3
Corneal neovascularisation Number of quadrants affected Follow‐up: 6 months to 24 months	450 per 1000	252 per 1000 (95 to 666)	RR 0.56 (0.21 to 1.48)	36 eyes (1 RCT)	⊕⊝⊝⊝ very low1,2,3
Symblepharon Presence of symblepharon Follow‐up: 6 months to 24 months	50 per 1000	21 per 1000 (1 to 474)	RR 0.41 (0.02 to 9.48)	36 eyes (1 RCT)	⊕⊝⊝⊝ very low1,2,3
Time‐to‐epithelialisation (days) Follow‐up: 6 months to 24 months	The mean time‐to‐epithelialization in the control group was 28.9	The mean time‐to‐epithelialization in the intervention group was 13 lower (26.30 lower to 0.30 higher)		36 eyes (1 RCT)	⊕⊝⊝⊝ very low1,2,3
Adverse effects	None reported (of 16 patients receiving AMT and 20 receiving standard medical therapy)			36 eyes (1 RCT)	⊕⊕⊝⊝ low1,3	The small sample makes it less likely to have rare but potentially important adverse effects
*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) AMT: amniotic membrane transplantation; CI: confidence interval; RR: risk ratio ** Visual acuity measured at final follow‐up rather than at a fixed interval.
Grading of Recommendations Assessment, Development and Evaluation (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.

¹Downgraded for risk of bias (‐1): high risk of performance and detection biases, as not possible to mask personnel and outcome assessors, and evidence of baseline imbalance in visual acuities between treatment arms. Epithelial defect assessed on day of clinic review, not daily. Defect was assessed under partially opaque membrane, suggesting possible imputation of data.

²Downgraded for imprecision (‐1): very wide confidence intervals including no effect. 

²Downgraded for imprecision (‐1): small study; it is likely the available evidence base is lower than the optimal information size.

Summary of findings 2. Amniotic membrane transplant (AMT) for severe ocular burns.

Amniotic membrane transplant (AMT) for severe ocular burns
Patient or population: people with severe acute ocular burns Settings: ophthalmic hospital Intervention: AMT and medical therapy Comparison: medical therapy alone
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of Participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	AMT
Failure of re‐epithelialisation by day 21 post‐injury  Image analysis of digital photographs Follow‐up: 6 months to 24 months	956 per 1000	985 per 1000 (899 to 1000)	RR 1.03  (0.94 to 1.12)	92 (2 RCTs)	⊕⊕⊝⊝ low1,2
**Visual acuity at final follow‐up LogMAR Scale from 0 to 3 (Lower values = better vision)  Follow‐up: 9 months to 24 months	The mean visual acuity at final follow‐up in the control groups ranged from  1.64 to 2.06	The mean visual acuity at final follow‐up in the intervention groups was 0.01 higher (worse) (0.29 lower to 0.31 higher)		92 eyes (2 RCTs)	⊕⊕⊝⊝ low1,2
Corneal neovascularisation Number of quadrants or central cornea involved Follow‐up: 9 months to 24 months	800 per 1000	672 per 1000 (528 to 848)	RR 0.84 (0.66 to 1.06)	92 eyes (2 RCTs)	⊕⊕⊝⊝ low1,2
Symblepharon Presence of symblepharon Follow‐up: 9 months to 24 months	444 per 1000	395 per 1000 (249 to 630)	RR 0.89 (0.56 to 1.42)	92 eyes (2 RCTs)	⊕⊕⊝⊝ low1,2
Time‐to‐epithelialisation (days) Follow‐up: 9 months to 24 months	The mean time‐to‐epithelialization in the control groups ranged from 40.6 to 75.8	The mean time‐to‐epithelialization in the intervention group was 1.66 lower (11.09 lower to 7.77 higher)		92 eyes (2 RCTs)	⊕⊕⊝⊝ low1,2
Adverse effects	None reported (of 47 patients receiving AMT and 45 receiving standard medical therapy)			92 eyes (2 RCTs)	⊕⊕⊝⊝ low1,2	Number of enrolled participants may be too low to detect rare adverse effects.
*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) ** Visual acuity measured at final follow‐up rather than at a fixed interval. AMT: amniotic membrane transplantation; CI: confidence interval; RR: risk ratio
Grading of Recommendations Assessment, Development and Evaluation (GRADE ) Working Group grades of evidence Grading of Recommendations Assessment, Development and Evaluation (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.

¹Downgraded for risk of bias (‐1): high risk of detection and performance biases related to difficulties in masking personnel and outcome assessors. Epithelial defect assessed on day of clinic review, not daily.

²Downgraded for imprecision (‐1): confidence intervals include clinically important benefit and harm. Sparse data (individual data not provided by one RCT).

Background

An ocular burn is potentially a sight‐threatening injury. The majority of the damage likely occurs within seconds of the initial injury. Mild burns can fully recover, but severe burns can permanently damage the ocular surface, leading to delayed epithelial healing, scarring of the conjunctiva, dry eye disease, and vascularisation and opacity of the cornea. In addition, there may be a secondary cataract, glaucoma, hypotony and phthisis (Bizrah 2019). After a burn, a key event that determines the likelihood of corneal recovery is the extent of damage to the corneal epithelial stem cells located in the basal layer of the epithelium at the limbus (the junction of the cornea and sclera). If stem cells are destroyed, they cannot regenerate, and there can be delayed recovery of an epithelial defect, with secondary overgrowth of the cornea's surface with an epithelium with a conjunctival phenotype (conjunctivalisation). The recovery of an intact epithelial layer can take weeks or months. Until the cornea has been re‐epithelialised, there is a risk of infection, corneal ulceration and corneal perforation. The treatment of an acute ocular burn is based on topical and oral medications. In addition, it has been proposed that an amniotic membrane placed on the surface of the eye in the acute phase, defined as the first seven days after the injury (McCulley 1987), will reduce pain, reduce the risk of corneal ulceration, encourage re‐epithelialisation of the cornea and conjunctiva, and preserve vision.

Description of the condition

Most ocular burns are caused by chemicals (acid, alkali) or heat (open flames, molten metal, fireworks). In population‐based surveys, between 7.3% and 22.1% (Wong 2000; Loon 2009; Wong 2018) of ocular injuries are caused by chemical burns, with a high incidence in males in low‐ and middle‐income countries associated with occupational exposure or assault (Kuckelkorn 1995a; Wagoner 1997, Xiang 2005; Macdonald 2009; Haring 2016). In children, injury from domestic cleaning products is common (Ratnapalan 2011; Vajpayee 2014). The immediate effect of an ocular burn is irreversible damage to ocular tissues, particularly the cornea and conjunctiva. Although alkaline agents penetrate the eye tissues faster than acids, the nature of the agent does not influence management beyond emergency treatment to remove debris and restore the pH of the tear film. The risk of delayed recovery and conjunctivalisation of the corneal surface following a burn is primarily determined by the amount of the circumference of the corneal limbus involved by the burn and the depth that the injury has penetrated beneath the surface. If tissue at the limbus is coagulated, there is usually associated destruction of the blood vessels, with vessel closure termed limbal ischaemia. Because the stem cells at the limbus cannot be directly visualised in vivo, vessel closure at the limbus, or an epithelial defect over the limbus, are used as proxy measurements to estimate limbal epithelial stem cell loss. A severe burn can also damage deeper structures such as the iris, lens, trabecular meshwork and ciliary body. Once the causative agent is removed, regenerative and reparative phases follow that can extend over many months. These treatment phases have been divided into acute (day 0 to 7), early reparative (day 7 to 21) and late reparative (McCulley 1987).

Limbal stem cell failure

The human corneal epithelial stem cells are located in a specialised environment (niche) that is essential to maintain their function within the basal layer of the corneal epithelium of the limbus, with the greatest concentration in the superior and inferior limbus (Davanger 1971; Kinoshita 1982; Cotsarelis 1989; Shortt 2007; Nowell 2017). The stem cells divide to maintain the corneal epithelium, which has a phenotype that is distinct from the conjunctival epithelium. Because of their relatively superficial location, the stem cells are susceptible to damage from an ocular burn. If the stem cells from a sector of the corneal limbus are lost, they cannot regenerate, and the more peripheral conjunctival epithelium migrates onto the cornea. If the overgrowth does not cross the visual axis, this does not affect vision. However, with more extensive damage to the circumference of the limbus, the conjunctival epithelium can cross the visual axis or cover the whole corneal surface, often accompanied by vascularisation and severe visual loss. Replacement of the corneal phenotype by conjunctiva is termed conjunctivalisation, and it is the result of limbal epithelial stem cell failure (Le 2018).

Classification of ocular burns

An assessment of the extent of damage to the surface of the eye in the acute phase can indicate the prognosis and final visual outcome. In 1964 Ballen published a classification of ocular burns categorised into four grades according to corneal opacity, chemosis, and ischaemia at the limbus with their associated prognosis (Ballen 1964). He noted that if > 50% of the limbus was ischaemic and the cornea opaque, the visual outcome was poor, with a guarded prognosis for retention of the globe. In 1965 Roper‐Hall published a classification of ocular burns similar to the Ballen classification but with emphasis on corneal opacity rather than chemosis (Table 3; Roper‐Hall 1965). The Roper‐Hall classification quantifies corneal epithelial loss, corneal opacity and the amount of the limbus affected by ischaemia. Corneal opacity was graded as 'hazy' or 'opaque' but with no assessment of the proportion of the cornea involved. There were four grades, concluding that the prognosis was poor if the cornea was opaque with more than 50% of the limbus ischaemic. Both these classifications predate the understanding of the role of limbal epithelial stem cells in maintaining the ocular surface.

1. Roper‐Hall classification.

Grade	Prognosis	Cornea	Conjunctiva
I	Good	Corneal epithelial damage	No limbal ischaemia
II	Good	Corneal haze, iris details visible	< 33% limbal ischaemia
III	Guarded	Total epithelial loss, stromal haze, iris details obscured	33% to 50% limbal ischaemia
IV	Poor	Cornea opaque, iris and pupil obscured	> 50% limbal ischaemia

In 2001 Dua and colleagues published an alternative to the Roper‐Hall classification (Table 4; Dua 2001). This was justified because advances in surgical rehabilitation, especially limbal epithelial stem cell transplantation, had meant that all severe burns that had > 50% limbal involvement did not have the same prognosis. The classification was based on the amount of epithelial defect involving the limbus or peripheral conjunctiva, demonstrated with fluorescein. Conjunctival staining at the limbus was counted in quadrants, while peripheral conjunctival staining was estimated in four groups of unequal size. The grading defined 50% to 75% of epithelial loss involving the limbus as having a 'good to guarded' prognosis, while 100% of epithelial loss had a very poor prognosis. Limbal ischaemia was excluded from the grading because it could not be reliably measured. The Roper‐Hall and the Dua grading systems were based on the authors' experience. Neither presented data to validate the classification or define the criteria for the grades of prognosis. In subsequent case series where both grading systems have been compared, there has been no clear superiority of either system (Westekemper 2017). However, in one analysis of data collected for a randomised controlled trial (RCT) of acute ocular burns (Tandon 2011), it was found that subdivision of the Dua classification grades IV to VI had predictive value for the risk of secondary corneal vascularisation, corneal clarity, and final visual acuity that was not apparent with the single grade IV of the Roper‐Hall classification (Gupta 2011). An assumption that there is an equivalence of the clinical signs that define the grades for the two systems has led to confusion when making comparisons (Gupta 2011; Tandon 2011; Sahay 2019). Finally, an assessment of epithelial defect or ischaemia made at the slit lamp or from photographs is subject to observer error (Kam 2019).

2. Dua classification.

Grade	Prognosis	Clinical findings	Conjunctival involvement	Analogue scale
I	Very good	0 clock hours of limbal involvement	0%	0/0%
II	Good	< 3 clock hours of limbal involvement	< 30%	0.1 to 3/1 to 29.9%
III	Good	3 to 6 clock hours of limbal involvement	> 30% to 50%	3.1 to 6/31 to 50%
IV	Good to guarded	6 to 9 clock hours of limbal involvement	> 50% to 75%	6.1 to 9/51 to 75%
V	Guarded to poor	9 to < 12 clock hours of limbal involvement	> 75% to < 100%	9.1 to 11.9/75.1 to 99.9%
VI	Very poor	Total limbus (12 clock hours) involved	Total conjunctiva (100%) involved	12/100%

*The analogue scale records the limbal involvement in clock hours of affected limbus/percentage of conjunctival involvement. While calculating percentage of conjunctival involvement, only involvement of bulbar conjunctiva, up to and including the conjunctival fornices is considered.

Two other systems for classification have been proposed, either based on an improved quantification of conjunctival stain with fluorescein (Harun 2004) or limbal ischaemia (Bagley 2006). Interestingly, the latter system proposes five grades of limbal ischaemia (no ischaemia, <2 5%, 25% to 50%, 50% to 75% and >75%). There is no reason that the different categories of severe limbal ischemia (50% to 75%, >75%) should not be incorporated into the Roper‐Hall classification. None of these systems includes parameters that indicate intraocular damage such as raised intraocular pressure, dilated iris, cataract, choroidal detachment, or associated lid injury and lagophthalmos. These may significantly impact the prognosis (Kuckelkorn 1995b).

Anterior segment optical coherence tomographic angiography (AS‐OCTA) has shown that a visual assessment of limbal ischaemia can underestimate the level of vessel closure and that conjunctival stain with fluorescein can significantly overestimate the extent of limbal ischaemia (Fung 2019; Ang 2021). Importantly, limbal stain with fluorescein is not a reliable index of limbal epithelial stem cell loss, and the circumferential extent of limbal ischaemia may change in the days following the acute assessment. Anterior segment angiography or AS‐OCTA can provide an objective and dynamic evaluation of the vasculature of the limbus that may become the standard to document the extent and depth of an acute injury for future clinical trials.

Description of the intervention

Medical treatment

Standard emergency medical treatment of an acute ocular burn includes removing any residual corrosive particulate matter and continuous irrigation with a phosphate‐free saline solution until the pH is neutralised (Bizrah 2019). Immediate treatment typically consists of a combination of a preservative‐free topical antibiotic and a corticosteroid. Additional topical treatment may also include a cycloplegic drop to relieve painful ciliary spasm. For severe burns, topical sodium ascorbate and potassium citrate drops are recommended to reduce the risk of corneal stromal ulceration and perforation. Systemic medications may include oral sodium ascorbate and tetracyclines, which may help reduce the risk of corneal stromal ulceration by inhibiting matrix metalloproteinases (Smith 2004). Oral acetazolamide may be necessary to control an acute rise in intraocular pressure.

Amniotic membrane transplantation

Some clinicians offer human amniotic membrane transplantation (AMT) as an adjunct to medical therapy for acute eye burns. The proposed benefits of AMT are to reduce pain, reduce inflammation, encourage healing and prevent complications. Treatment consists of an onlay of the membrane that is sutured or glued to the margins of the lids to cover the whole of the conjunctival and corneal surfaces (Tamhane 2005). Less frequently, it is sutured to the cornea at the limbus to cover just the cornea or used as a combined corneal patch and conjunctival onlay. Variations include supporting the membrane with a plastic ring or silicone conformer or using a disc of an amniotic membrane supported by a soft contact lens. The membrane is not, in reality, a transplant as there is no intention for the membrane to be integrated into the ocular tissue, and over time it dissolves or is dislodged. It should therefore be considered as a patch or protective bandage. The tissue is usually cryopreserved or freeze‐dried before use, and no living cells are transferred. No common adverse effects have been reported from using an amniotic membrane patch. An AMT can also be used as a substrate for the late reconstruction of the eye with limbal epithelial cells or tissue, which is not the subject of this review (Shimazaki 1997; Le 2019).

How the intervention might work

Physical properties of amniotic membrane

Amniotic membrane is a thin, pliable tissue of approximately 50 μm thickness. There is an epithelial and a stromal surface, but the orientation of the amniotic membrane when used as a patch is not proven to be clinically important. Using a sheet of amniotic membrane to cover an ocular surface following a burn may reduce pain (Shafto 1950; Arora 2005), possibly by physically protecting the inflamed ocular surface whilst allowing oxygen transfer to occur (Baum 2002). The mechanical separation of the inflamed conjunctival surfaces may also reduce the risk of symblepharon formation but is not thought to prevent conjunctivalisation following severe burns (Meller 2000).

Biological properties of amniotic membrane

In addition to its physical properties, it has been proposed that the amniotic membrane may also act as a biomaterial to promote epithelial healing and suppress inflammation (Kim 2000; Tseng 2004; Liu 2012; Mamede 2012). It contains a range of cytokines that some think has the potential to reduce inflammation and angiogenesis (Paolin 2016). However, it is unknown how long these cytokines persist in the tissue after transplantation or whether they have a clinically significant benefit.

Alternative surgical management of acute ocular burns

Alternative surgical approaches to protect the ocular surface include advancing the vascularised subconjunctival Tenon's layer anteriorly to the limbus (Tenon‐plasty) (Kuckelkorn 1995b). Tenon‐plasty can only be carried out if there is sufficient residual viable and vascularised tissue. Temporary tarsorrhaphy is a further option to protect the ocular surface. A normal corneal epithelial phenotype can be restored by stem cell transplantation, but this type of surgery is unlikely to succeed in the early healing phase whilst there is active inflammation (Rao 1999).

Why it is important to do this review

Acute ocular burns vary in severity and prognosis. Mild burns can heal completely with only minimal intervention. Very severe burns can lead to loss of the eye despite treatment. Between these two extremes, there is a range of severity where intervention may affect outcomes. AMT is widely used as a treatment for acute ocular burns, although there is a lack of certainty as to its effectiveness. Case series have demonstrated either a positive effect or a lack of effect. Ophthalmic surgeons need to know how they should treat the various grades of burn and whether this treatment will improve the outcome.

Objectives

To compare the effect of : amniotic membrane transplantation (AMT) with medical therapy in the first seven days after an ocular surface burn, compared to medical therapy alone.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) reported in any language irrespective of their publication status.

Types of participants

Study participants of any age who had an acute chemical or thermal ocular burn of severity grade II or worse. We did not apply any restrictions regarding location, setting or demographic factors. We excluded participants who received amniotic membrane (AM) treatment after the acute phase. We considered the following two populations separately.

• Moderate burns ‐ patients with ocular burns of Grades II and III (Roper‐Hall and Dua classifications).

• Severe burns ‐ patients with ocular burns of Grade IV (Roper‐Hall)/Grades IV to VI (Dua classification).

Types of interventions

• We included trials that used one or more AM patches to cover the cornea and the whole or part of the ocular surface to the lid margins, applied within seven days of injury, and combined with standard medical therapy. We excluded patients treated after the seven‐day window. We excluded trials that compared AM extracts or suspensions with any other treatment.

• The control patients received standard treatment, with no surgical intervention other than manual lysis of adhesions between the eyelids and the globe.

Types of outcome measures

We did not select particular outcomes as a criterion for eligibility or exclude studies if an outcome of interest was not reported. We collected data on primary and secondary outcome measures.

Primary outcomes

• The proportion of eyes with failure of epithelialisation 21 days after the burn injury.

• The visual acuities (logMAR) at final follow‐up.

Secondary outcomes

• The proportion of eyes with symblepharon.

• The proportion of eyes with new vessels in the cornea (corneal neovascularisation).

• Time‐to‐epithelialisation.

• The proportion of eyes with adverse events.

Follow‐up

For the primary outcomes, the assessment was stipulated as 21 days after the injury because an epithelial defect at the end of the early repair phase carries a risk of subsequent corneal complications. A difference between treatment and control groups at this time point is therefore likely to be clinically significant. Attrition from loss to follow‐up at day 21 is likely to be low, and it is unlikely that other secondary surgeries would have been performed. The minimum length of follow‐up was stipulated as six months.

Search methods for identification of studies

Electronic searches

The Cochrane Eyes and Vision Information Specialist developed the searches to identify randomised controlled trials and controlled clinical trials. However, during the development of the review and subsequent updates, the searches were run without a filter to retrieve reports of any study design. There were no restrictions to language or year of publication. The search was last updated on 29 September 2021.

• Cochrane Central Register of Controlled Trials (CENTRAL; 2022, Issue 9) (which contains the Cochrane Eyes and Vision Trials Register) in the Cochrane Library (searched 29 September 2021) (Appendix 1).

• MEDLINE Ovid (January 1946 to 29 September 2021) (Appendix 2).

• Embase Ovid (January 1980 to 29 September 2021) (Appendix 3).

• Latin American and Caribbean Health Sciences (LILACS) (January 1982 to 29 September 2021) (Appendix 4).

• ISRCTN registry (www.isrctn.com/editAdvancedSearch; searched 29 September 2021)

• US National Institute of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; searched 29 September 2021).

• World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/search/en; searched 29 September 2021).

Searching other resources

We searched the references lists of included studies for other possible studies. We contacted the authors of included studies to find out if there were ongoing trials.

Data collection and analysis

Selection of studies

Two review authors working independently (GC, ST) screened titles and abstracts of all articles identified by the search and independently assessed full‐text reports of all potentially eligible studies. If abstracts were not available, full‐text articles were screened. In the event of a disagreement regarding eligibility, CB would arbitrate. For any full‐text reports that were rejected, the reasons were recorded in the Characteristics of excluded studies table.

Data extraction and management

Where possible, we extracted data onto a data extraction form to obtain details of methodology and outcomes. Two review authors (GC, ST) extracted the study characteristics and one review author (GC) checked the data before entering it into RevMan Web (RevMan Web 2022). In the event of disagreement, CB arbitrated. General information entered included demographic data, diagnostic criteria and the grades and causative agent of the injury, and the timing and method of application of AM. We used the form to confirm eligibility for the review. In the event of missing data, we contacted the trial authors for more details. For dichotomous outcomes, we entered the proportion of patients in each intervention arm having the outcome of interest. We recorded continuous outcomes with all relevant statistical data.

Assessment of risk of bias in included studies

We conducted a domain‐based evaluation of the risk of bias within each study, using an assessment tool as described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). The tool includes six domains: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Within each domain, assessments were made of relevant outcomes, and concise descriptions of the relevant characteristics were added. Where possible, information was obtained from published trial reports and contacts with the investigators. When no information was available, this was stated. Judgements of high, low, or unclear risk of material bias were then assigned for each item. Two review authors (GC, ST) independently assessed the risk of bias in each domain (see risk of bias table in 'Characteristics of included studies'). Any discrepancies were resolved by discussion in the first instance, followed by arbitration by the third review author when necessary.

Measures of treatment effect

Measures of effect included the risk ratio (RR) for dichotomous outcomes (failure of epithelialisation, corneal neovascularisation, symblepharon and adverse events) and mean difference (MD) for continuous variables (logMAR visual acuity and time to epithelialisation).

Unit of analysis issues

Eyes and people

We assessed whether studies had included one or two eyes from each participant and whether randomisation was conducted at the level of the participant or the eye. If one eye was chosen for bilateral patients, we documented the method of selection in order to detect possible bias. Analyses based on a single eye per individual conveniently allow standard statistical methods to be employed, although information for the fellow eye is lost (Murdoch 1998).

If both eyes were included, and treatment to one eye influenced the treatment to the other, we considered this as a source of bias. If both eyes were treated similarly, the likely correlation between fellow eyes was assumed to result in a possible cluster effect. Therefore, we considered analyses based on both injured eyes permissible if the proportion of bilateral eye injuries was small (less than 1%).

Dealing with missing data

We assessed whether there were patients for whom outcomes were not assessed. In the case of missing outcome data, we aimed to compare the characteristics of patients with missing data with those of patients with complete data to detect possible bias. We checked whether intention‐to‐treat analyses had been conducted, as a failure to do so could introduce bias.

Assessment of heterogeneity

We assessed heterogeneity between studies by review of study characteristics. We assessed statistical heterogeneity in effect estimates across studies using the Chi² test (P = 0.10). We assessed inconsistency between studies by examination of the I² statistic with confidence intervals (CIs), using 75% as a cut‐off.

Assessment of reporting biases

To mitigate publication bias, we extended our search to unpublished studies and trial registers. We did not impose any language restrictions. We intended to use a funnel plot to assess the risk of publication bias if there were more than 10 trials within our review.

Data synthesis

To investigate differential outcomes of AMT we performed separate analyses on moderate and severe categories of acute burns. In the event of two or more suitable RCTs being identified we planned to conduct a meta‐analysis where the data appeared clinically, methodologically, and statistically homogeneous. We used a fixed‐effect model due to the sparsity of the data available. This model, which assumes that the effect size is identical for all studies, was chosen because although it ignores potential heterogeneity in the evidence base, it may be expected to provide more robust estimates of the average intervention effect in this situation.

Subgroup analysis and investigation of heterogeneity

We did not plan or perform subgroup analyses, since moderate and severe burns can have very different natural histories and can therefore be regarded as separate categories.

Sensitivity analysis

We planned to conduct sensitivity analyses based on the risk of bias (low, unclear, high) concerning the reported treatment effect on each outcome measure, but were unable to proceed due to not having enough RCTs.

Summary of findings and assessment of the certainty of the evidence

We consulted Chapter 11 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011) for the completion of a summary of findings table of included studies. We considered six outcomes of interest for moderate burns (Table 1) and severe burns (Table 2).

• The proportion of eyes with failure of epithelialisation by day 21 after the injury

• The visual acuity at final follow‐up

• Corneal neovascularisation

• Symblepharon

• Time‐to‐epithelialisation

• Adverse effects

Results

Description of studies

Results of the search

The original electronic searches run in June 2012 identified 104 records. After duplicates were removed we screened 88 records and rejected 80 records as not eligible for inclusion in the review. We obtained full‐text copies of eight reports for further assessment. We excluded five prospective studies (Arora 2005; López‐García 2006; López‐García 2007; Meller 2000; Muraine 2001) and one randomised study (Tamhane 2005) (see 'Characteristics of excluded studies' for reasons for exclusion)).

An updated search run in September 2021 identified 265 references (Figure 1). After 89 duplicates were removed, we removed a further 174 records that were not relevant to the scope of the review. We obtained full‐text copies of two reports for further assessment.We included  Eslani 2019  in this update of the review and excluded Sharma 2016.

1.

Flow diagram

Included studies

In this update we included two RCTs (Tandon 2011; Eslani 2019). In Tandon 2011, a subset of patients (n = 68) treated within the first seven days of the injury met the inclusion criteria. The subset consisted of 36 moderate (Roper‐Hall classification II‐III) and 32 severe (Roper‐Hall classification IV) ocular burns. In the treatment group, a single amniotic membrane patch was secured at the corneal limbus and the lid margins. A second RCT (Eslani 2019) looked exclusively at 60 severely injured eyes (Roper‐Hall Grade IV) of 60 patients. In the treatment group of this RCT, one amnion patch was first sutured onto the cornea, with a second patch that covered the entire ocular surface; any residual amniotic membrane (AM) was removed at three weeks.

In Tandon 2011 males and females of any age were treated within seven days of a chemical or thermal burn to one or both eyes. A subset of 68 eyes from 100 eyes recruited was treated in the first seven days following a burn and met the inclusion criteria (Clare 2012). These data were provided by the study authors following direct correspondence. Patients were randomised to receive cryopreserved amniotic membrane transplantation (AMT) and standard medical therapy or standard medical therapy alone. The primary outcome measure was the rate of healing of the corneal epithelial defect. The study authors provided information on the proportion of eyes with incomplete epithelialisation by day 21 post‐injury.

In Eslani 2019 males and females of any age who were seen within seven days of a chemical or thermal burn to one or both eyes. Nine out of 69 participants (seven treated after the initial seven‐day window, two lost to follow‐up) were excluded, the remaining 60 eyes met the study inclusion criteria. Patients were randomised to receive cryopreserved AMT and standard medical therapy or standard medical therapy alone. The primary outcome measure was time to complete corneal epithelialisation. The authors did not respond to a request for additional data.

No missing data were reported in either study.

See the Characteristics of Included Studies for more information.

Ongoing studies

We did not identify any ongoing studies.

Excluded studies

We excluded five non‐randomised prospective studies (Arora 2005; López‐García 2006; López‐García 2007; Meller 2000; Muraine 2001). Two RCTs were also excluded.

• One RCT was not stratified into acute and early phase groups, with insufficient data on the number of eyes with intact epithelium at 21 days and visual acuity (Tamhane 2005). The study also randomised 19% of the eyes from bilaterally injured patients. However, experience gained from this study was used by the same centre to design a second RCT by Tandon and colleagues, which has been included (Tandon 2011).

• One RCT (Sharma 2016) did not meet the minimum criteria for length of patient follow‐up or dealing with unit‐of‐analysis issues as set out in the Cochrane protocol  for this review (Clare 2011), and could not be used for further analysis. The study also randomised 66% of eyes from bilaterally injured patients, and follow‐up was limited to three months. These protocol deviations precluded the study from a further meta‐analysis. It is unclear whether there was an overlap in patient recruitment from a previous retrospective case series (Sharma 2015).

Risk of bias in included studies

We used the Cochrane Collaboration's tool for assessing the risk of bias for two outcomes: epithelial closure and final visual acuity (Characteristics of included studies). The risk of bias is summarised across domains for both study‐level entries such as allocation, and outcome‐specific entries such as masking. A Risk of Bias summary for both studies is shown in Figure 2. The study protocol highlights the importance of the unit‐of‐analysis, stipulating a maximum threshold of 1% bilateral injuries (Clare 2011).

2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Each RCT took a different approach to cases of bilateral burns. In Tandon 2011, the right eye was selected, while Eslani 2019 selected the more severely injured eye.

Tandon 2011:the patients were randomised to a treatment assignment list prepared from a random numbers table. Serial numbers were given to the cases, and the randomly allocated treatment decision was concealed by using sealed envelopes. This suggests that adequate measures were taken to prevent the foresight of treatment at the point of enrolment. However, serious baseline imbalances exist which are unexplained (unclear). For example, close inspection of the data shows that three eyes (out of 20) in the moderate burns group took 90 days to heal, suggesting a higher degree of severity. These three eyes were allocated to the control group. By contrast, the AMT group had only one eye with a comparable healing time (60 days). In addition, there was a significant difference in baseline visual acuity between the treatment and control groups in the moderate burns category. The mean logMAR visual acuity at presentation was 0.45 in the treatment group (standard deviation (SD) 0.29) and 0.92 in the controls (SD 0.88). This could have skewed the final visual outcomes in favour of a treatment effect. As there was no obvious reason for this discrepancy, the allocation bias was downgraded (high).

Eslani 2019: Randomisation was performed according to the random block permutation method, and the study allocation was concealed to the investigators by using sealed envelopes (low).

Blinding

Tandon 2011: it was not possible to mask study participants and personnel from knowing which treatment was received, suggesting a high risk of performance bias (unequal care). Although masked observers assessed outcomes using digital photographs, the assessors would know whether AMT had been allocated, indicating a high risk of detection bias (Higgins 2017) (high).

Eslani 2019:it was not possible to mask study participants and personnel from knowing which treatment had been received, suggesting an unclear risk of performance bias. The detection bias is unclear because the AMT was removed at three weeks (unclear).

Incomplete outcome data

Tandon 2011: no participants were lost to follow‐up. Data on epithelial healing rates may have been incomplete for two reasons. Firstly, in some cases, the margins of the epithelial defect could not be identified through the amniotic membrane, yet there is no indication that data were missing or excluded, and it is not explained how the problem was resolved. Second, the epithelial defects were not measured daily, suggesting that the precise timing of epithelial closure was an estimate, and that data were imputed (Higgins 2017) (high).

Eslani 2019:tTwo participants were lost to follow‐up, presumably after randomisation. This could have influenced the results, indicating an unclear risk of attrition bias (unclear). The time‐to‐epithelialisation is given in days, but the patients were not reviewed on a daily basis suggesting that data were imputed (high).

Selective reporting

The authors of the Tandon 2011 study make it clear from the outset that they have confidence in the value of AMT, raising concerns about their impartiality and, therefore, in their reporting. The improvement in vision measured before and after treatment was greater in the control eyes (AMT ‐0.39, 95% confidence interval (CI) ‐0.23 to ‐0.55 versus controls ‐0.54, 95% CI ‐0.08 to ‐1.00). The implication that AMT improved final visual acuity is therefore misleading. In addition, biased reporting may have occurred if the patients treated within seven days of injury had different characteristics from patients treated outside the seven‐day window (high).

In Eslani 2019, the possibility of a selective reporting bias in the study was unclear. Although the study includes 60 eyes of 60 patients with Roper‐Hall Grade IV burns, further descriptive data are not shown (unclear).

Other potential sources of bias

The risk of bias is shown in brackets for each potential source of bias. Cases in which the direction of bias favoured a treatment effect were graded as having a high risk of bias.

Tandon 2011

• There was a longer follow‐up for treated eyes (mean difference (MD) 14.5 months, range 10 to 24) in the moderate category than for control eyes (MD 12 months, range 6 to 24). Since there was no common time‐point for reporting visual acuity, this constitutes a potential source of bias (unclear).

• Of the 68 patients included in the study, 24 subsequently had secondary procedures, such as limbal stem cell transplantation, following the initial injury. It is unclear whether these were eyes that had suffered more serious injuries (unclear).

Eslani 2019

• Seven patients were excluded because they presented >7 days after the injury. It is not clear whether exclusion occurred before or after randomisation. The injuries in these cases may have differed in severity from the included patients (unclear).

• It is impossible to assess whether the distribution of severity of the injuries in the two groups is equal (unclear).

Effects of interventions

See: Table 1; Table 2

Moderate burns (Roper‐Hall Grades II‐III)

Failure of epithelialisation by day 21 post‐injury

Seven out of 20 control eyes and one out of 16 treatment eyes had failed to epithelialise by 21 days, a relative risk of 0.18 in the treatment group (95% confidence interval (CI) 0.02 to 1.31). The lack of masking, the baseline imbalance between treatment and control groups and wide confidence interval justified downgrading the GRADE assessment to very low.

Visual acuity

At final follow‐up, the mean visual acuity (LogMAR) was 0.06 in the AMT treatment group (SD 0.10). In the control group, LogMAR visual acuity was 0.38 (SD 0.52). The mean difference (MD) was ‐0.32 (95% CI ‐0.09 to ‐0.55). Lack of masking and baseline imbalance led us to downgrade the GRADE assessment to low.

Corneal neovascularisation

The RR of developing corneal new vessels in the AMT treatment group was 0.56 (4/16 eyes in the treatment group and 9/20 eyes in the control group) (95% CI 0.21 to 1.48). The GRADE assessment was downgraded to very low as a result of the baseline imbalance and the wide confidence intervals.

Symblepharon

The RR of developing symblepharon among the AMT participants in the moderate group was 0.41 (0/16 treatment and 1/20 control eyes) (95% CI 0.02 to 9.48). Division of adhesions was performed in the control group but not the AMT treatment group, which may have affected the outcome. The GRADE assessment was downgraded to very low as a result of the baseline imbalance and the wide confidence intervals.

Time‐to‐epithelialisation

In the control group, time‐to‐epithelialisation was 28.9 (SD 27.1 days; range 7 to 60 days), whereas in the AMT intervention group it was 15.9 (SD12.2 days; range 8 to 90 days), which is 13 days lower (26.30 lower to 0.30 higher). The data are represented by a survival curve (Figure 3). Lack of masking, serious baseline imbalance, measurement through an opaque membrane and non‐daily follow‐up suggesting possible imputation of data led us to downgrade the GRADE assessment to very low.

3.

Survival curves of epithelial defects of moderate ocular burns treated during first 7 days with AMT (data obtained from authors of Tandon 2011).

Adverse events

Of the 16 patients recorded as having received an AMT, none reported any adverse events. GRADE assessment was judged to be low due to the risk of bias.

Severe burns (Roper‐Hall Grade IV)

Failure of epithelialisation by day 21 post‐injury

Sixteen of 17 eyes treated with AMT and 14/15 control eyes (Tandon 2011), and 30/30 AMT eyes and 29/30 control eyes (Eslani 2019) had failed to epithelialise by 21 days. This gives a RR of 1.03 (95% CI 0.94 to 1.12). GRADE assessment was judged to be low due to risk of bias and imprecision.

Visual acuity

At final follow‐up, the mean visual acuity (LogMAR) in the control groups was 1.64 (SD 1.48) (Tandon 2011) and 2.06 (SD 0.57) (Eslani 2019). In the intervention groups, respective visual acuity measurements were 1.77 (SD 1.31) and 2.06 (SD 0.67). Combining the data gives a mean difference (MD) of 0.01 (95% CI ‐0.29 to +0.31). GRADE assessment was judged to be low due to risk of bias and imprecision.GRADE assessment was judged to be low due to risk of bias and imprecision.

Corneal neovascularisation

The RR of developing corneal new vessels in the combined AMT intervention groups was 0.84 (32/47 eyes in the treatment groups and 36/45 eyes in the control groups) (95% CI 0.66 to 1.06).  GRADE assessment was judged to be low due to risk of bias and imprecision.

Symblepharon

The RR of developing symblepharon among the AMT participants in the severe group was 0.89 (19/47 treatment and 20/45 control eyes) (95% CI 0.56 to 1.42). Division of adhesions was performed in the control group but not the AMT treatment group, which may have affected the outcome. GRADE assessment was judged to be low due to risk of bias and imprecision.

Time‐to‐epithelialisation

The data for severe burns from one RCT (Tandon 2011) are represented by a survival curve (Figure 4). In the control group, time‐to‐epithelialisation was 40.6 (SD 17.1 days; range 12 to 80 days), whereas, in the AMT intervention group, it was 39.9 (17.5 days; range 14 to 80 days). Individual data on time‐to‐epithelialisation was not provided on request to the authors of Eslani 2019, precluding survival curve analysis. For this RCT, time‐to‐epithelialisation in the control group was 75.8 (SD 25.8 days; range 46 to 170 days), whereas in the AMT intervention group it was 72.6 (SD 32.4 days; range 21 to 180 days). The combined mean time‐to‐epithelialisation was 1.66 days lower (11.09 lower to 7.7 higher). Confidence intervals include both clinically important benefit and harm, indicating imprecision. Follow‐up was non‐daily, suggesting possible imputation of data. GRADE assessment was downgraded to low.

4.

Survival curves of epithelial defects of severe ocular burns treated during first 7 days with AMT (data obtained from authors of Tandon 2011).

Adverse events

Of the 47 patients recorded as having received an AMT, none reported any adverse events. GRADE assessment was judged to be low due to the risk of bias and small sample size.

Discussion

Summary of main results

Determinants of outcome

An acute burn to the eye can cause rapid and permanent damage to the ocular surface, depending on the severity. Variations in the causative agent, length of exposure, delay in presentation and treatment contribute to the high level of outcome variability. Following a mild burn, there may be a complete visual recovery with minimal intervention, while a very severe burn may cause loss of all vision despite treatment. If > 5 0% of the limbus is affected by the injury, there is a high risk of delayed epithelial closure, resulting in overgrowth of epithelium with a conjunctival phenotype onto the cornea (conjunctivalisation). It has been proposed that covering the ocular surface in the acute phase with an amniotic membrane may encourage healthy corneal re‐epithelialisation and improve outcomes. This approach has been incorporated in several recent published protocols for managing severe burns (Westekemper 2017; Sharma 2018; Bizrah 2019; Dua 2020). Because the severity of the burn is the primary determinant of the outcome, this must be accurately assessed and graded at recruitment for any controlled study. Two commonly used grading systems (Roper‐Hall 1965; Dua 2001) suffer from poor interobserver repeatability (Kam 2019; Ang 2021) and a poor correlation between the observed clinical signs and an objective assessment of limbal ischaemia (Fung 2019; Ang 2021). Introducing techniques such as anterior segment optical coherence tomographic angiography (AS‐OCTA) that can document both the depth and extent of vascular closure may provide a more accurate and consistent grading for future studies.

Description of outcome measures

The two primary outcome measures addressed in the review are the proportion of eyes with a residual epithelial defect (i.e. failure of epithelialisation) at the end of the early reparative phase (day 21) and the final visual acuity. Secondary outcome measures include corneal neovascularisation, symblepharon formation, time‐to‐epithelialisation and adverse events.

Evidence from randomised controlled trials (RCTs)

The Tandon 2011 RCT recruited 100 participants. In the moderate burns group (Grade II‐III Roper‐Hall classification); the authors reported significantly faster epithelial healing in the amniotic membrane transplantation (AMT) treatment arm and better visual outcomes, suggesting a benefit of AMT in this category. There was no evidence of significant benefit from intervention with AMT in the severe category (Grade IV Roper‐Hall classification). By analysing a smaller subset of patients from this RCT, we obtained numerically different results from the original study and consequently changes in statistical significance. Moreover, our critical appraisal of the study and application of GRADE led us to a divergent interpretation of the results.

A further RCT on severe burns (Eslani 2019) found no difference in time‐to‐epithelialisation, final visual acuity, symblepharon formation and corneal neovascularisation between treatment and intervention groups. The injuries sustained were classified as Grade IV (Roper‐Hall classification), which covers a spectrum of severity from 50% to 100% limbal ischaemia.

The two RCTs do not show a consistent advantage in epithelial healing, final visual acuity, symblepharon formation and corneal neovascularisation in eyes having a supplementary AMT.

Overall completeness and applicability of evidence

We could only include two RCTs in this systematic review, indicating a need for further studies. We assessed failure of epithelialisation by day 21, final visual outcomes, corneal neovascularisation, symblepharon formation, time‐to‐epithelialisation and adverse effects.

The RCT by Tandon 2011 looked at both moderate and severe injuries, and included patients treated outside the seven day 'acute' window. The authors reported faster epithelial healing in eyes treated with AMT, but no difference in final visual outcomes, corneal neovascularisation or symblepharon formation. Individual data were provided to us by the study authors, allowing us to analyse the subset treated during the acute phase and extract the necessary data for our analysis.

In contrast to Tandon 2011, Eslani 2019 focused exclusively on severely injured eyes, potentially missing any benefits AMT might have conferred for moderate burns. The outcome measures included time‐to‐epithelialisation, final visual outcomes, corneal neovascularisation and symblepharon formation. The authors did not respond to our requests for individual data.

Overall, we found a significant risk of bias due to lack of masking and baseline imbalance, as well as serious imprecision in measurement, wide confidence intervals and small sample size. The risk of bias across several domains is summarised in Figure 2. This caused us to downgrade the certainty of the evidence as to whether amniotic membrane transplantation (AMT) in the first seven days after an ocular surface burn can improve any of the clinical outcomes, limiting the applicability of the RCTs.

In addition, the evidence base is incomplete. For example, while It has been claimed that AMT may have a role in reducing pain in the acute phase after an injury, this has not been assessed in any formal RCT, and therefore could not be included as an outcome in this review. 

The value of this systematic review is to highlight the lack of certainty supporting the use of AMT in acute ocular burns and to provide a template for future RCTs.

Quality of the evidence

We evaluated the certainty of evidence in the Tandon 2011 and Eslani 2019 RCTs for the two primary outcomes and four secondary outcomes, using GRADE criteria.

The wide confidence intervals in both primary and secondary outcome measures imply imprecision. Further imprecision was suggested by non‐daily follow‐up, implying possible data imputation. Moreover, while complete descriptive data from the Tandon 2011 study was made available to us, no data were forthcoming from the Eslani 2019 study.

Together with the risk of bias due to lack of masking and baseline imbalances, these factors led us to downgrade the certainty to very low for five outcomes in the moderate burns category (Table 1) and to low (failure of epithelialisation, visual acuity, corneal neovascularisation and symblepharon) and very low (time‐to‐epithelialisation) in the severe burns category (Table 2).

Potential biases in the review process

We have included all relevant studies in this review. We based the decision to exclude one RCT (Tamhane 2005) on the difficulty of obtaining all the necessary data. We excluded a more recent RCT (Sharma 2016) because of the duration of inadequate follow‐up and unit‐of‐analysis issues, which could have influenced outcomes. These exclusions may be a source of bias. Additionally, choosing a fixed‐effect model could be seen as a source of bias, but was considered more robust because of sparsity of data. In practice, however, the differences between fixed‐effect and random‐effects calculations were negligible, such that choice of model would not affect the conclusions.

Agreements and disagreements with other studies or reviews

We identified one other systematic review with meta‐analysis, which also considered the effect of eyelid surgery (Klifto 2019). Based on the single report of Sharma and colleagues, they concluded that the use of an AMT significantly increased the proportion of fully healed epithelial defects at three months following an ocular burn (Sharma 2016).

Authors' conclusions

Implications for practice.

This review has identified randomised controlled (RCT) data highlighting the lack of certainty to support amniotic membrane transplantation (AMT) for the treatment of acute ocular burns, either to improve the chance of corneal re‐epithelialisation or to improve the visual outcome. In the spectrum of severity of the injury, mild ocular burns have a good prognosis, and AMT is not usually indicated. In contrast, there is no evidence that an AMT will mitigate the blinding sequelae for very severe burns. Between these extremes, there may be a subgroup of eyes where an AMT may benefit, but a more accurate case definition would be required to identify suitable cases for future trials. Overall, these data should be considered when clinicians and other decision‐makers discuss the management options for acute ocular burns.

Implications for research.

This review demonstrates that an RCT can be performed to assess the effectiveness and safety of an AMT as a treatment option for the management of an acute ocular burn. Although clinical outcome measures are essential, a future study design should consider including additional outcome measures of health‐related quality of life and other aspects important to people who have had severe ocular injuries. This should include parameters that determine the suitability of the eye for future reconstructive surgery (limbal transplantation, keratoplasty).

This review has also highlighted some practical considerations relevant to future RCTs.

• Classifying the severity of an acute ocular burn is problematic. Both the Roper‐Hall and the Dua classification are imprecise tools subject to observer error, with a lack of validation. The role of anterior segment optical coherence tomographic angiography (AS‐OCTA) as an objective technique for characterising the severity of ocular surface burns should be defined.

• Ocular burns are uncommon, and a multi‐centre design may be necessary to achieve adequate study power.

• Adequate follow‐up is essential to allow the full consequences of the burn to become apparent

What's new

Date	Event	Description
8 June 2021	New search has been performed	Search updated
23 August 2020	New citation required but conclusions have not changed	New RCT included

History

Protocol first published: Issue 10, 2011
Review first published: Issue 9, 2012

Date	Event	Description
17 October 2013	Amended	Contact details updated.

Appendices

Appendix 1. CENTRAL search strategy

#1 MeSH descriptor Eye Burns
#2 (eye* or ocular) near/6 (burn*)
#3 (#1 OR #2)
#4 MeSH descriptor Amnion
#5 amniotic near/3 membrane*
#6 AMT
#7 (#4 OR #5 OR #6)
#8 (#3 AND #7)

Appendix 2. MEDLINE Ovid search strategy

1. randomized controlled trial.pt.
2. (randomized or randomised).ab,ti.
3. placebo.ab,ti.
4. dt.fs.
5. randomly.ab,ti.
6. trial.ab,ti.
7. groups.ab,ti.
8. or/1‐7
9. exp animals/
10. exp humans/
11. 9 not (9 and 10)
12. 8 not 11
13. Eye burns/
14. ((eye$ or ocular) adj6 burn$).tw.
15. or/13‐14
16. exp Amnion/
17. (amniotic adj3 membrane$).tw.
18. AMT.tw.
19. or/16‐18
20. 15 and 19
The search filter for trials at the beginning of the MEDLINE strategy is from the published paper by Glanville 2006.

Appendix 3. Embase Ovid search strategy

1. exp randomized controlled trial/
2. exp randomization/
3. exp double blind procedure/
4. exp single blind procedure/
5. random$.tw.
6. or/1‐5
7. (animal or animal experiment).sh.
8. human.sh.
9. 7 and 8
10. 7 not 9
11. 6 not 10
12. exp clinical trial/
13. (clin$ adj3 trial$).tw.
14. ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$)).tw.
15. exp placebo/
16. placebo$.tw.
17. random$.tw.
18. exp experimental design/
19. exp crossover procedure/
20. exp control group/
21. exp latin square design/
22. or/12‐21
23. 22 not 10
24. 23 not 11
25. exp comparative study/
26. exp evaluation/
27. exp prospective study/
28. (control$ or prospectiv$ or volunteer$).tw.
29. or/25‐28
30. 29 not 10
31. 30 not (11 or 23)
32. 11 or 24 or 31
33. Eye burn/
34. Cornea burn/
35. ((eye$ or ocular) adj6 burn$).tw.
36. or/33‐35
37. exp Amnion/
38. (amniotic adj3 membrane$).tw.
39. AMT.tw.
40. or/37‐39
41. 36 and 40

Appendix 4. LILACS search strategy

amniotic and eye or ocular burn$

Appendix 5. ISRCTN search strategy

amniotic and ocular burn

Appendix 6. ClinicalTrials.gov search strategy

amniotic AND ocular burn

Appendix 7. WHO ICTRP search strategy

amniotic AND ocular burn

Data and analyses

Comparison 1. AMT with medical therapy versus medical therapy alone ‐ moderate burns.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Failure of epithelialisation (21 days)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1.1 Moderate burns	1	36	Risk Ratio (M‐H, Fixed, 95% CI)	0.18 [0.02, 1.31]
1.2 Visual acuity at final follow‐up	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.2.1 Moderate burns	1	36	Mean Difference (IV, Fixed, 95% CI)	‐0.32 [‐0.55, ‐0.09]
1.3 Corneal neovascularisation	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.3.1 Moderate burns	1	36	Risk Ratio (M‐H, Fixed, 95% CI)	0.56 [0.21, 1.48]
1.4 Symblepharon	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.4.1 Moderate burns	1	36	Risk Ratio (M‐H, Fixed, 95% CI)	0.41 [0.02, 9.48]
1.5 Time‐to‐epithelialisation	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.5.1 Moderate burns	1	36	Mean Difference (IV, Fixed, 95% CI)	‐13.00 [‐26.30, 0.30]
1.6 Adverse effects	1		Other data	No numeric data
1.6.1 Moderate burns	1		Other data	No numeric data

1.1. Analysis.

Comparison 1: AMT with medical therapy versus medical therapy alone ‐ moderate burns, Outcome 1: Failure of epithelialisation (21 days)

1.2. Analysis.

Comparison 1: AMT with medical therapy versus medical therapy alone ‐ moderate burns, Outcome 2: Visual acuity at final follow‐up

1.3. Analysis.

Comparison 1: AMT with medical therapy versus medical therapy alone ‐ moderate burns, Outcome 3: Corneal neovascularisation

1.4. Analysis.

Comparison 1: AMT with medical therapy versus medical therapy alone ‐ moderate burns, Outcome 4: Symblepharon

1.5. Analysis.

Comparison 1: AMT with medical therapy versus medical therapy alone ‐ moderate burns, Outcome 5: Time‐to‐epithelialisation

1.6. Analysis.

Comparison 1: AMT with medical therapy versus medical therapy alone ‐ moderate burns, Outcome 6: Adverse effects

Adverse effects
Study	AMT events	AMT total	Medical therapy events	Medical therapy total
Moderate burns
Tandon 2011	0	16	0	20

Comparison 2. AMT with medical therapy versus medical therapy alone ‐ severe burns.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Failure of epithelialisation (21 days)	2	92	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.94, 1.12]
2.2 Visual acuity at final follow‐up	2	92	Mean Difference (IV, Fixed, 95% CI)	0.01 [‐0.29, 0.31]
2.3 Corneal neovascularisation	2	92	Risk Ratio (M‐H, Fixed, 95% CI)	0.84 [0.66, 1.06]
2.4 Symblepharon	2	92	Risk Ratio (M‐H, Fixed, 95% CI)	0.89 [0.56, 1.42]
2.5 Time‐to‐epithelialisation	2	92	Mean Difference (IV, Fixed, 95% CI)	‐1.66 [‐11.09, 7.77]
2.6 Adverse effects	2		Other data	No numeric data

2.1. Analysis.

Comparison 2: AMT with medical therapy versus medical therapy alone ‐ severe burns, Outcome 1: Failure of epithelialisation (21 days)

2.2. Analysis.

Comparison 2: AMT with medical therapy versus medical therapy alone ‐ severe burns, Outcome 2: Visual acuity at final follow‐up

2.3. Analysis.

Comparison 2: AMT with medical therapy versus medical therapy alone ‐ severe burns, Outcome 3: Corneal neovascularisation

2.4. Analysis.

Comparison 2: AMT with medical therapy versus medical therapy alone ‐ severe burns, Outcome 4: Symblepharon

2.5. Analysis.

Comparison 2: AMT with medical therapy versus medical therapy alone ‐ severe burns, Outcome 5: Time‐to‐epithelialisation

2.6. Analysis.

Comparison 2: AMT with medical therapy versus medical therapy alone ‐ severe burns, Outcome 6: Adverse effects

Adverse effects
Study	AMT events	AMT total	Medical therapy events	Medical therapy total
Eslani 2019	0	30	0	30
Tandon 2011	0	17	0	15

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Eslani 2019.

Study characteristics
Methods	Study design: Prospective, randomised, single centre, controlled, interventional study
Participants	Country: Iran Total number of participants: 60 (60 eyes). 69 patients enrolled. In bilateral burns, the more severely affected eye was randomised. Number (%) of men and women: 56 (93%) male and 4 (7%) female Age range: 25 years (SD 7; range 12 to 49) Review: Participants were seen on Days 1, 3, 7, 14 and 28 Then every two weeks until 3 months Then monthly until 1 year Then quarterly Follow‐up: minimum 12 months mean 20.3 months (SD 2.5; range 13‐24) Type of burn: All had Grade IV (Roper Hall classification) 27 (45%) acid, 33 (55%) alkaline Inclusion criteria: Grade IV burns seen within 7 days of injury Exclusion criteria: Exclusion from recruitment not reported Seen >7 days from injury (7 patients) Loss to follow‐up (2 patients) Standard medical therapy: Topical preservative‐free lubricating gel and drops, chloramphenicol (0.5% four times per day), betamethasone (0.1% 2 hourly), homatropine (2% three times per day) Oral vitamin C (500 mg  four times per day) and doxycycline (100 mg twice per day) Topical and systemic anti‐glaucoma medications as required
Interventions	Intervention: One amniotic membrane was secured at the limbus and 3 mm beyond the limbus (two 10/0 nylon sutures), with a second amniotic membrane to cover the whole ocular surface and attached at the lid margin (6/0 vicryl sutures). Residual membrane removed at 3 weeks Standard medical therapy Division of adhesions Comparator: Standard medical therapy Division of adhesions
Outcomes	Primary outcome: Time to complete corneal epithelialisation Secondary outcomes: Best‐corrected visual acuity Neovascularisation in the central 5 mm of the cornea Complications (symblepharon, glaucoma) No data were provided on the reduction of pain or the proportion of eyes with epithelial stem cell failure (conjunctivalisation)
Notes	Date conducted: August 2006 to January 2014 Sample size calculation: to have an 80% power for detecting a difference of 14 days in the mean corneal epithelial healing time between the groups as significant (for the 2‐sided 5% level) with an assumed SD of 20 days, 29 eyes in each group were required. Trial ID: NCT00370812 Sources of funding: this work was partially supported by R01 EY024349 (ARD) and Core Grant for Vision Research P30 EY001792 from NEI/NIH; Vision for Tomorrow (ARD), and an unrestricted grant to the department from Research to Prevent Blindness (all located in the US). The funders had no role in study design, data collection and analysis, decision to publish, or manuscript preparation. Declaration of interests: one author acknowledged financial support from other sources for research, speaking honoraria, and consultation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The random block permutation method was used.
Allocation concealment (selection bias)	Low risk	Concealed to investigators via opaque sealed envelopes.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	The presence of the AMT would be evident. It would not be possible to mask participants and personnel from knowing which treatment was received.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	The presence of the AMT would be evident to the assessor until its removal after Day 21. All clinical examinations were done by one of the authors.
Incomplete outcome data (attrition bias) All outcomes	High risk	Unclear if the exclusion of 2 patients lost to follow‐up occurred before or after randomisation. These losses may have influenced the results. Data on the precise timing of epithelial closure may have been imputed.
Selective reporting (reporting bias)	Unclear risk	Results were not reported based on a possible treatment effect. However, data were not made available for this review, precluding verification.
Other bias	Unclear risk	Grade IV (Roper‐Hall) ocular burns include limbal ischaemia ranging from 50% to 100%. There is weak evidence that a burn with 50% ischaemia has a better prognosis than one with 100% ischaemia, particularly for corneal vascularisation, a secondary outcome measure for this study. A benefit for intervention for milder burns within the Grade IV group may have been missed. In addition, the study population includes children, but the proportions of children in the two groups are not presented. Epithelial healing in children may differ from adults.

Tandon 2011.

Study characteristics
Methods	Study design: prospective, randomised, single centre, single surgeon, controlled, interventional study
Participants	Country: New Delhi, India Total number of participants: 100 patients were enrolled, of which 68 fulfilled the inclusion criteria for this review (seen within seven days of injury). Bilateral cases were randomised as individuals, and the right eye was selected for randomisation. Number (%) of men and women: Moderate burn group ‐ 34 (94%) males and 2 (6%) females Severe burn group ‐ 24 (75%) males and 8 (25%) females Age range: Moderate burn group ‐ mean age of 22.3 years [SD 10.1 (median 20.5, range 4 to 52) Severe burn group ‐ mean age of 19.3 years (SD 15.8; median 12.5, range 6 to 61) Review: patients were seen on Days 1 and 7 Then weekly until 1 month Then every 2 weeks until three months Monthly after that Follow‐up: Range 6 to 24 months Type of burn: Moderate burn group (Roper‐Hall Grades II‐III) ‐ 22 burns were caused by alkali, 10 by acid, and 4 by direct heat Severe burn group (Roper‐Hall Grade IV) ‐ 23 burns were caused by alkali, 7 by acid, and 2 by direct heat The study used a modified grading of corneal clarity. Anterior segment angiography was performed, although the results were not reported. Inclusion criteria: Grades II‐IV burn Acute ocular burn seen within 7 days Exclusion criteria: Criteria for exclusion from recruitment not reported Seen >7 days from injury (not reported) Loss to follow‐up (0) Standard medical therapy: Topical prednisolone acetate (1%) every 6 hours, ofloxacin (0.3%) every 6 hours, sodium ascorbate (10%) every 4 hours, sodium citrate (10%) every 4 hours and preservative‐free tear substitutes every 2 hours; homatropine (2%) twice daily Oral vitamin C (500 mg) every 6 hours for 2‐4 weeks Antiglaucoma therapy, including timolol maleate 0.5% drops and/or oral acetazolamide as required
Interventions	Intervention: Amniotic membrane patch sutured to entire ocular surface within seven days of injury. Peri‐limbal and lid margin 8/0 Vicryl sutures were used to secure the membrane Conventional medical therapy Comparator: Conventional medical therapy Mechanical release of adhesions
Outcomes	Primary outcome: Rate of healing of corneal epithelial defect Number of eyes with completely corneal re‐epithelialised at 21 days Secondary outcome: Visual acuity at final follow‐up Corneal clarity Number of quadrants of corneal vascularisation Presence of symblepharon Adverse events
Notes	Date conducted: October 2003 to December 2005 Sample size calculation: There was no power calculation, and stratified random sampling was performed to ensure equal patient numbers in the two burn severity groups. The sample size of 100 cases was calculated based on the expected number of patients who could be recruited during a 2‐year study period based on past clinic and emergency records. The authors provided a database. The results presented here constitute a secondary analysis that differs from the published version. Trial ID: CTRI/2009/091/001018 Source of funding: Indian Council of Medical Research, Ansari Nagar, New Delhi under Grant‐in‐aid of Research Scheme (No I‐389) Declaration of interests: none declared
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The patients were randomised using a treatment assignment list prepared with the help of a table of random numbers. Comment: Probably adequate, but severe baseline imbalances exist which are unexplained and favour a treatment effect.
Allocation concealment (selection bias)	High risk	Serial numbers were given to the cases, and concealed randomisation using sealed envelopes was followed to decide treatment. Comment: It was probably adequate, although there was no explicit mention of when the envelopes were opened or if they were opaque. Severe baseline imbalances are unexplained and could have skewed results in favour of treatment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	The presence of AMT is evident to participants and clinicians, with a potential for unequal care. Not possible to mask treatment.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Digital photographs at each visit were obtained and stored for independent comparative assessment by masked observers. Comment: The presence of the AMT would be evident on the digital photographs, at least for a few weeks following surgery, implying possible unequal assessment.
Incomplete outcome data (attrition bias) All outcomes	High risk	There are no reports of incomplete follow‐up. Some of the outcome data on epithelial healing may have been imputed.
Selective reporting (reporting bias)	High risk	The authors state in the Introduction that 'AMT improves corneal re‐epithelialisation and hastens recovery after chemical burns. Amniotic membrane is effective in reducing inflammation and its consequences, and may partially restore limbal stem cell function in ocular burns'. Comment: potential reporting bias.
Other bias	Unclear risk	There are differences in the length of follow‐up between the treatment and control groups. A subset of patients had secondary procedures. The study population includes children, but the proportions in the two groups are not presented.

SD: standard deviation

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Arora 2005	Prospective non‐randomised study
López‐García 2006	Prospective non‐randomised study
López‐García 2007	Prospective non‐randomised study
Meller 2000	Prospective non‐randomised study
Muraine 2001	Prospective non‐randomised study; non‐acute injuries
Sharma 2016	Randomisation of 66% of the eyes from bilaterally injured patients. Each eye was randomised separately, thus ignoring the possible patient‐specific factors on outcomes Bilateral burns formed the majority of cases, leading to possible cluster effects. Study follow‐up was limited to three months, which was shorter than the protocol specified period of six months
Tamhane 2005	In eyes with severe burns, there was a significant baseline imbalance with the AMT treatment group having more severe injuries (significantly larger epithelial defects and worse visual acuity) and a high rate of participant dropout The randomisation strategy was inadequate Some of the interventions were performed after seven days. Seven (19%) bilateral cases were also included, which could lead to cluster effects. There were insufficient data on the number of eyes with intact epithelium at 21 days and on visual acuity. The authors were contacted and they confirmed that the required data were no longer available.

AMT: amniotic membrane transplantation

Differences between protocol and review

Instead of using the pre‐specified review outcome "the proportion of eyes with complete epithelialisation" 21 days after a burn injury, we used the proportion of eyes with "failure of epithelialisation" (i.e. a persistent epithelial defect at day 21). This was because it is the more clinically significant finding as a risk factor for corneal melting and a poor outcome, and merits further analysis.

The randomised controlled trials (RCTs) did not report on the proportion of eyes with a fibrovascular pannus (a cause of disfigurement and visual loss) or on pain reduction following : amniotic membrane transplantation (AMT).

Because we have included only two RCTs, we could not produce a funnel plot comparing different treatment effects. We have not conducted sensitivity analyses for the main findings.

The search strategies published in the protocol did not include an RCT filter as we were considering searching for all study designs. However, after further discussion, we incorporated an RCT filter as the review will only use data from RCTs.

Contributions of authors

• Conceiving the review: GC

• Designing the review: GC

• Co‐ordinating the review: GC

• Data collection for the review:

  • Designing electronic search strategies: Cochrane Eyes and Vision Group editorial base

  • Undertaking manual searches: GC, ST

  • Screening search results: GC, ST

  • Organising retrieval of papers: GC

  • Screening retrieved papers against inclusion criteria: GC, ST, CB

  • Appraising quality of papers: GC, ST, CB

  • Extracting data from papers: GC, ST, CB

  • Writing to authors of papers for additional information: GC

  • Providing additional data about papers: GC

  • Obtaining and screening data on unpublished studies: GC

• Data management for the review:

  • Entering data into Review Manager: GC

  • Checking that data entered into Review Manager is correct: GC

• Analysis of data: GC and CB

• Interpretation of data: GC

  • Providing a methodological perspective: GC, CB

  • Providing a clinical perspective: GC, ST

  • Providing a policy perspective: GC, ST, CB

• Writing the review: GC, ST

• Providing general advice on the review: CB, ST, CB

• Securing funding for the review

• Performing previous work that was the foundation of the current study: GC, HSD, HS, CB

Sources of support

Internal sources

• New Source of support, UK

  None

External sources

• Public Health Agency, UK

  The HSC Research and Development (R&D) Division of the Public Health Agency funds the Cochrane Eyes and Vision (CEV) editorial base at Queen's University Belfast. 

• Queen's University Belfast, UK

  Gianni Virgili, Co‐ordinating Editor for Cochrane Eyes and Vision’s work, is funded by the Centre for Public Health, Queen’s University Belfast, Northern Ireland.

Declarations of interest

GC: None known.
CB: None known.
ST: None known.

New search for studies and content updated (no change to conclusions)