Search tips
Search criteria 


Logo of brjopthalBritish Journal of OphthalmologyVisit this articleSubmit a manuscriptReceive email alertsContact usBMJ
Br J Ophthalmol. 2007 August; 91(8): 1048–1053.
Published online 2007 February 21. doi:  10.1136/bjo.2006.109124
PMCID: PMC1954786

The natural history of Stevens–Johnson syndrome: patterns of chronic ocular disease and the role of systemic immunosuppressive therapy



To characterize patterns of chronic ocular disease in patients with Stevens–Johnson syndrome (SJS) and its variant toxic epidermal necrolysis (TEN), and to describe their response to treatment.


Retrospective case series. A review of hospital records of 30 patients (60 eyes) with ocular manifestations of SJS or TEN was carried out. The principal outcome measure was to identify and classify the patterns of chronic ocular disease in SJS and TEN. The secondary outcome measure was the response to treatment.


Patterns of chronic ocular disease observed after the acute episode included: mild/moderate SJS, severe SJS, ocular surface failure (SJS‐OSF), recurrent episodic inflammation (SJS‐RI), scleritis (SJS‐S) and progressive conjunctival cicatrisation resembling mucous membrane pemphigoid (SJS‐MMP). The median follow‐up was 5 years (range 0–29). 19 patients (29 eyes (48%)) developed SJS‐OSF, SJS‐RI, SJS‐S or SJS‐MMP during follow‐up. SJS‐OSF was present in 12 patients (18 eyes (30%)). In 5 patients (eight eyes) this developed 1 year after the acute illness, without any further inflammatory episodes; it was associated with SJS‐RI in 1 patient (2 eyes), with SJS‐RI and SJS‐S in 1 patient (1 eye), with SJS‐S in 1 patient (1 eye) and with SJS‐MMP in 4 patients (6 eyes). Episodes of SJS‐RI occurred in 4 patients (7 eyes (12%)). The median time from acute disease to the first episode of SJS‐RI was 8.5 years (range 5–63). SJS‐S developed in 2 patients (4 eyes (7%)), of which 2 eyes subsequently developed SJS‐OSF. SJS‐MMP developed in 5 patients (10 eyes (16.6%)). The median duration from the acute stage to the diagnosis of SJS‐MMP was 2 years (range 1–14). Immunosuppressive therapy successfully controlled inflammation in 10/10 patients with SJS‐MMP, SJS‐RI or SJS‐S.


Ocular disease in SJS/TEN is not limited solely to the sequelae of the acute phase illness. Patients and physicians need to know that ocular disease progression, due to surface failure and/or acute inflammatory conditions, may occur at variable periods following the acute disease episode. Recognition of this, and prompt access to specialist services, may optimise management of these uncommon patterns of disease in SJS.

Erythema multiforme major (Stevens–Johnson syndrome) (SJS) is a complex immunological syndrome characterised by acute blistering affecting the skin and at least two mucous membranes. Toxic epidermal necrolysis (TEN) is the most severe form of erythema multiforme and involves >30% of the skin, which sloughs in sheets.

Although severe corneal damage can result from the effects of the acute stage of the disease, it more usually develops as a result of the chronic complications. Corneal damage is the most severe long‐term complication for survivors of TEN and SJS.1,2,3,4,5

Little information exists regarding the natural history of chronic ocular disease in SJS. The chronic complications have different aetiologies and may co‐exist or occur sequentially in the same patient.6 Whereas a subset of patients have mild ocular involvement, other patients present with severe complications early after the acute stage, caused by conjunctival scarring.1,2,3,4,5 A major cause of chronic disease is stem cell failure (ocular surface failure in SJS (SJS‐OSF)).7 Another cause of late sight‐threatening complications is recurrent conjunctival inflammation (SJS with recurrent inflammation (SJS‐RI)),8 not associated with trichiasis, entropion, keratoconjunctivitis sicca or lid margin keratinisation. An ocular mucous membrane pemphigoid picture has also been described in some cases of late SJS (SJS with ocular mucous membrane pemphigoid (SJS‐MMP)).9 Late diffuse scleritis complicating SJS has also been reported (SJS‐S).10 As the management strategies differ for these conditions, identification of these infrequently described patterns of late exacerbations in SJS is important to the successful management of this condition.

In this study we identify these different causes of chronic eye involvement in SJS, and the timing of their development, in order to describe the natural history of chronic eye disease in SJS. The outcome of therapies used to manage these causes of chronic eye involvement is described.

Patients and methods

All patients attending Moorfields Eye Hospital (MEH) from 1975 to 2004 with SJS or TEN were identified from the hospital's electronic patient records and from the Cornea and External Disease Service immunosuppression database. The inclusion criterion was patients with a dermatological diagnosis of SJS or TEN and eye involvement. As the ocular manifestations of these diseases are similar,5 they are not differentiated in the analysis, and both conditions are referred to as SJS.

The principal outcome measure was the classification of the causes of chronic ocular involvement in SJS. The secondary outcome was the response of these to treatment. For the principal outcome, the different causes of ocular disease were classified using nomenclature from previous reports and, where our patients developed different disease variants, this has been described. For each patient, each eye was assigned to one of these conditions and, if another cause of disease developed, the date in relation to the onset of disease was recorded. The following classification of disease in SJS was used (fig 1):

  • Mild/moderate SJS1,2,3,4,5,6: mild/moderate sequelae resulting from conjunctival scarring. These were eyes with healthy corneas, no recurrent or progressive inflammation and no progressive conjunctival scarring.
  • Severe SJS1,2,3,4,5,6: severe sequelae resulting mainly from conjunctival scarring. These included eyes with a clinical diagnosis of ocular surface disease, without definitive evidence of stem cell deficiency. Eyes were very dry with an opaque, vascularised and irregular surface, but with a stable epithelium and no evidence of conjunctival inflammation or progressive scarring.
  • Ocular surface failure11 (SJS‐OSF): surface failure with evidence of stem cell deficiency occurring after the acute episode of SJS. Dysfunction of the stem cells of the corneal epithelium was identified by late fluorescein staining, superficial vascularisation, signs of complete conjunctivalisation or keratinisation of the cornea, an unstable epithelium or cytological evidence of stem cell failure (presence of goblet cells on the corneal surface, and absence of cytokeratin CK3 and presence of cytokeratin CK19 on immunohistochemistry).12
  • Recurrent inflammation8 (SJS‐RI): eyes with recurrent episodes of conjunctival inflammation, occurring in the absence of either external aggravating factors or other non‐ocular mucosal manifestations of recurrent disease, and without evidence of progressive scarring.
  • Scleritis10 (SJS‐S): patients developing scleritis after the acute episode of SJS.
  • Mucous membrane pemphigoid (SJS‐MMP): eyes with early or late9 conjunctival inflammation, not associated with external aggravating factors (such as lid margin disease), and leading to progressive scarring with or without basement membrane antibodies on biopsies.

This study was conducted after approval by the local research governance committee.


Thirty patients (10 male and 20 female; median age 38 years (range 12–79)) who had developed SJS (24) or TEN (6) and who had been treated at MEH were included. Drugs were the most common triggering factor (18 cases), with antibiotics implicated in seven cases. An infectious aetiology was identified in nine cases and the precipitating agent was unknown in four of these cases. In two patients, there was both a history of previous drug intake and infection. Immunisation was the precipitating agent in one case. The median follow‐up was 5 years (range 0–29). The median age at onset of the initial acute disease was 18 years (range 1–69). The median time from the acute episode to referral was 3 years (range 0–53). Table 11 summarises the study population, their patterns of chronic ocular disease and immunosuppressive treatment used.

Table thumbnail
Table 1 Summary of the study population and patterns of chronic ocular disease

Mild/moderate SJS was diagnosed in four patients (cases 1–4, eight eyes) at presentation and remained stable throughout follow up.

Severe SJS was present in seven patients (cases 5–11), in at least one eye, at presentation and remained stable throughout their follow‐up.

Severe SJS developed in four patients (cases 12–15), who initially had mild/moderate SJS, but whose disease progressed, in one or both eyes, as a consequence of the chronic effects of surface disease without late acute episodes of inflammation.inflammation.

figure bj109124.f1
Figure 1 Natural history of progression of ocular disease following onset of SJS/TEN.
figure bj109124.f3
Figure 3 Right eye of case 27 showing progressive conjunctival cicatrisation (SJS‐MMP).

SJS‐OSF developed in 12 patients (18 eyes) (cases 16–27). Five of these patients (eight eyes, cases 16–20) developed late surface failure, without any acute episodes of inflammation, between 1 and 2 years after the onset of the acute episode of SJS/TEN; cases 17 and 20 (three eyes) had had minimal corneal disease immediately after recovery from the acute phase of SJS/TEN. Three patients (four eyes, cases 21–23) developed SJS‐OSF following inflammation due to SJS‐RI (case 21), both SJS‐RI and SJS‐S (case 22), and SJS‐S (case 23 and fig. 22).

figure bj109124.f2
Figure 2 Right eye of case 22 showing scleritis (SJS‐S).

SJS‐RI episodes developed in four patients (seven eyes) during their follow‐up, cases 21 and 22 described above who also developed SJS‐OSF and cases 29 and 30 who had no other late complications of the disease. Three of these eyes (cases 29 and 30) had been classified as mild/moderate SJS when seen for the first time at MEH. The time from the acute disease to the first episode of recurrent inflammation in these cases was 5, 8, 63 and 9 years, respectively (median 8.5 years).

SJS‐MMP developed in five patients (10 eyes) with either mild/moderate or severe SJS (cases 24–28 and fig.3). The median time from the acute stage to the diagnosis of SJS‐MMP was 2 years (range 1–14 yesrs). In four cases (24–27) (six eyes) SJS‐OSF also developed.

Conjunctival biopsy, for routine histopathology and direct immunofluorescence, was carried out in seven patients (cases 4, 13, 16, 22, 25, 27 and 30). Biopsy results were normal in five patients. In case 27 there was an inflammatory infiltrate along the epithelial–stromal margin, and immunofluorescence showed immunoglobulin (Ig)M, IgG and C3 staining at the conjunctival basement membrane, consistent with the development of ocular mucous membrane pemphigoid. In case 13, the conjunctival biopsy showed granular complement deposition, consistent with SJS‐associated conjunctivitis.8

Immunosuppression was initiated with both systemic corticosteroids and a steroid‐sparing immunosuppressive agent. The corticosteroids were tapered and stopped once the inflammation was brought under control. Immunosuppressive therapy successfully controlled inflammation in 10 patients (see table 11).). In case 4, the patient had been prescribed oral prednisolone and dapsone in the first year after the acute episode, before the first visit to MEH; however, there had been little or no response. When she first attended MEH her disease was classified as mild and she did not require immunosuppression. Immunosuppressive therapy was needed in another three patients to prevent corneal graft rejection (cases 8 and 13) and as complementary treatment of persistent epithelial defect after superficial keratectomy (case 18). Immunosuppressive drugs given included: prednisolone (12 patients), intravenous methylprednisolone (2 patients), dapsone (4 patients), sulphapyridine (1 patient), azathioprine (5 patients), mycophenolate (5 patients), adalimumab (1 patient), ciclosporin (5 patients), infliximab (2 patients) and cyclophosphamide (1 patient).

Apart from two patients, in whom two or more steroid‐sparing agents were used simultaneously (cases 26 and 22), the majority were treated with one drug. In some cases, patients required changes in treatment (9 patients), due to either intolerance or inadequate control of inflammation. In one patient (case 30), systemic immunosuppression was not tolerated, so it was discontinued and control of inflammation was achieved with topical corticosteroids. In case 27, dapsone and sulphapyridine were not tolerated, and two recent episodes of inflammation were controlled with short courses of oral prednisolone.

Visual acuity did not change in 16 eyes, increased in 19 eyes and decreased in 21 eyes (table 11).

The individual case reports are given in Appendix 1 available on the BJO website at


SJS and TEN are the most severe variants of the spectrum of erythema multiforme, an acute immunological mucocutaneous disorder with high morbidity and mortality, and an incidence of 1.89 per million inhabitants per year.13 Approximately 80% of hospitalised patients develop acute ocular complications that are similar for both SJS and TEN, with severe involvement in 25%.4,5 The morbidity of the disease may be due to the acute corneal complications, but is more usually due to the results of conjunctival scarring,6 with chronic sequelae occurring in approximately 35% of patients.14

The severity of the consequences of conjunctival scarring range from mild to severe. Some patients in this series presented with mild ocular involvement, caused by mild dry eye or meibomian gland disease, and showed no progression towards the more severe stages. Others presented with severe complications early after the acute stage. However, it is our experience, and that of other authors,6,8,9 that chronic ocular complications after SJS/TEN can present with a variable course that is not always the direct consequence of conjunctival scarring.

Late surface failure (SJS‐OSF) may occur in SJS/TEN both without late onset inflammatory disease and also as a consequence of three types of late inflammatory disease: SJS‐RI, SJS‐S and progressive conjunctival scarring (SJS‐MMP).

Surface failure (SJS‐OSF) can appear early, as a consequence of severe inflammation in the acute stage of SJS, or can manifest as late as 4 years after the initial onset of the disease.7 Destruction of the limbus during the acute phase of the disease may cause stem cell deficiency, but leave enough transient amplifying cells to maintain a normal corneal epithelium for some time after the disease onset; this could explain the evolution from peripheral, or localised, surface failure to complete surface failure within 2 years, which developed in five cases (eight eyes) in this series. Similar findings have been described in one case of SJS and in patients with thermal/chemical injury.7 Late stem cell failure may also appear as a result of prolonged limbal inflammation, in cases with recurrent SJS, SJS‐MMP and SJS‐S, as occurred in eight patients (10 eyes) in this report. For this reason, we strongly recommend aggressive management of any inflammatory episodes.

SJS‐RI has been reported in a small group of SJS patients with recurrent conjunctival inflammation unrelated to mechanical irritation or to dryness, but perhaps linked to immune complex vasculitis, and not associated with cutaneous relapses.8 In all these patients, the epithelial basement membrane was negative for immunoreactant deposition. The occurrence of scleritis in one patient with SJS has been reported.10 SJS‐MMP has been described in five cases following an acute episode of severe ocular inflammation after SJS.9 The time lag between the onset of SJS and MMP ranged from a few months to 31 years. All five patients had linear immune deposits characteristic of cicatricial pemphigoid along the basement membrane zone of a mucosal biopsy; all responded to systemic immunosuppressive therapy.

We present four cases (seven eyes) of SJS‐RI and five cases (10 eyes) of SJS ‐MMP. Only one (case 27) of the two patients with SJS‐MMP who underwent conjunctival biopsy had immunoglobulin deposition along the basement membrane. However, positive direct immunofluorescence is present in only 50% of patients with ocular MMP,15 so that this investigation is only helpful when positive.16 For this reason, four patients were diagnosed as SJS‐MMP, following the development of progressive conjunctival scarring, but in the absence of positive immunofluorescence. Progressive cicatricial changes were absent both in our patients with SJS‐RI and in the previously described cases.8 Conjunctival biopsies in two of our cases of SJS‐RI (22, 30) did not show perivasculitis.8 SJS‐S occurred in two patients (four eyes) and was severe, with limbitis leading to stem cell deficiency in one eye. Robin et al10 have described histopathological findings in a conjunctival–scleral biopsy, similar to those which Foster reported in the conjunctiva of SJS‐RI patients8; the development of bilateral necrotising sclerokeratitis, after conjunctival inflammation (case 22), could point to a common pathophysiological pathway for both patterns of chronic ocular disease.

As many of these patients were tertiary referrals, we cannot make any statement about the incidence of these different conditions. However, the handful of previous reports suggest these are very infrequent.7,8,9,10

The management of the chronic stage of SJS/TEN is one of the most challenging problems for the ophthalmologist. Treatment is directed at the management of the ocular surface, eliminating or minimising treatment toxicity and suppressing inflammation.6 Successful management requires identification of these different components of the disease, which are not present in all cases. Conjunctival inflammation in SJS may have a non‐specific component related to external factors such as trichiasis or severe dry eye24 so that endogenous inflammation cannot be assessed until the external triggering factors of conjunctival inflammation have been controlled; the residual inflammation then reflects the activity of endogenous inflammation in those cases with recurrent or persistent inflammation. It is in these cases that the use of systemic immunosuppression with corticosteroids and/or steroid‐sparing agents must be considered; topical corticosteroids are rarely useful in the long term for these cases.8,9,10 Treatment guidelines are similar to those for autoimmune bullous disorders except that the use of sulphonamides should be avoided, as these can precipitate recurrences of SJS.6,15 If patients have conjunctival inflammation or evidence of progressive conjunctival shrinkage, cataract surgery or oculoplastic surgery on the conjunctiva (as opposed to lid surgery for trichiasis and distichiasis) should be avoided until the disease is controlled by systemic immunosuppressive therapy.16

This review of patients with SJS/TEN illustrates several different causes of delayed ocular disease (SJS‐OSF, SJS‐RI, SJS‐S and SJS‐MMP) that may occur in the natural history of the disease that follows the acute episode of SJS/TEN. These are not recognised by many ophthalmologists because these events have been infrequently reported and are not all described in reviews or in text books. Their recognition is relevant, both to the prognosis for vision and so that appropriate anti‐inflammatory therapy can be instituted.

Supplementary Material

[web only appendix]


The authors are grateful to Suzanne Cabral and the Research Department at Moorfields Eye Hospital for logistical support, and to Sister Melanie Phillips for assistance in identifying the patients.


MEH - Moorfields Eye Hospital

MMP - mucous membrane pemphigoid

OSF - ocular surface failure

RI - recurrent inflammation

S - scleritis

SJS - Stevens–Johnson syndrome

TEN - toxic epidermal necrolysis


MV De Rojas was supported by Sociedad Gallega de Oftalmologia, Spain.

Competing interests: None.


1. Holland E J, Hardten D R. Stevens–Johnson syndrome. In: Pepose JS, Holland GN, Wilhelmus KR, eds. Ocular infection & immunity. St Louis, Missouri: Mosby, 1996. 416–425.425
2. Robin H, Hoang‐Xuan T. Fibrosing conjunctivitis. In: Hoang‐Xuan T, Baudouin Ch, Creuzot‐Garcher C, eds. Inflammatory diseases of the conjunctiva. Stuttgart: Thieme, 2001. 73–95.95
3. Mondino B J. Cicatricial pemphigoid and erythema multiforme. Ophthalmology 1990. 97939–952.952 [PubMed]
4. Power W J, Ghoraishi M, Merayo‐Lloves J. et al Analysis of the acute ophthalmic manifestations of the erythema multiforme/Stevens–Johnson syndrome/toxic epidermal necrolysis disease spectrum. Ophthalmology 1995. 1021669–1676.1676 [PubMed]
5. Pleyer U, Haberle H, Baatz H. et al Acute manifestations of oculo‐muco‐cutaneous disorders: erythema multiforme major, Stevens–Johnson syndrome, and toxic epidermal necrolysis. In: Pleyer U, Hartmann Chr, Sterry W, eds. Oculodermal diseases. Buren, The Netherlands: Aeolus Press, 1997. 169–191.191
6. Leonard J N, Dart J K G. The skin and the eyes. In: Burns, ed. Rooke's dermatology. Oxford: Blackwell Scientific, 2004. 64.1–
7. Puangsricharern V, Tseng S C G. Cytologic evidence of corneal diseases with limbal stem cell deficiency. Ophthalmology 1995. 1021476–1485.1485 [PubMed]
8. Foster C S, Fong L P, Azar D. et al Episodic conjunctival inflammation after Stevens–Johnson syndrome. Ophthalmology 1988. 95453–462.462 [PubMed]
9. Chan L S, Soong H K, Foster C S. et al Ocular cicatricial pemphigoid occurring as a sequela of Stevens–Johnson syndrome. JAMA 1991. 2661543–1546.1546 [PubMed]
10. Robin H, Hoang‐Xuan T, Bouder P. et al Le syndrome de Stevens–Johnson, une nouvelle cause de sclerite? Bull Soc Ophthalmol Fr 1993. 9665–672.672
11. Dua H S, Joseph A, Shanmuganathan V A. et al Stem cell differentiation and the effects of deficiency. Eye 2003. 17877–885.885 [PubMed]
12. Elder M J, Hiscott P, Dart J K. Intermediate filament expression by normal and diseased human corneal epithelium. Hum Pathol 1997. 28(12)1348–1354.1354 [PubMed]
13. Rzany B, Mockenhaupt M, Baur S. et al Epidemiology of erythema exsudativum multiforme majus, Stevens–Johnson syndrome and toxic epidermal necrolysis in Germany (1990–1992): structure and results of a population‐based registry. J Clin Epidemiol 1996. 49769–773.773 [PubMed]
14. Yetiv J Z, Bianche J R, Owen J A. Etiologic factors of the Stevens–Johnson syndrome. South Med J 1980. 73599–602.602 [PubMed]
15. Bernauer W, Elder M J, Leonard J. et al The value of biopsies in the evaluation of chronic progressive conjunctival cicatrisation. Graefe's Arch Clin Exp Ophthalmol 1994. 23533–537.537 [PubMed]
16. Bernauer W. Chronic progressive forms of cicatrising conjunctivitis. In: Pleyer U, Hartmann Chr, Sterry W, eds. Oculodermal diseases. Buren, The Netherlands: Aeolus Press, 1997. 193–221.221
17. Dart J. Corneal toxicity: the epithelium and stroma in iatrogenic and factitious disease. Eye 2003. 17886–892.892 [PubMed]
18. Collin J R O. Entropion and trichiasis. In: Collin JRO, ed. A manual of systematic eyelid surgery, 2nd Edn. London: Churchill Livingstone, 1989. 8–26.26
19. Kaido M, Goto E, Dogru M. et al Punctal occlusion in the management of chronic Stevens–Johnson syndrome. Ophthalmology 2004. 111895–900.900 [PubMed]
20. Poon A C, Geerling G, Dart J K G. et al Autologous serum eyedrops for dry eyes and epithelial defects: clinical and in vitro toxicity studies. Br J Ophthalmol 2001. 851188–1197.1197 [PMC free article] [PubMed]
21. Noble B A, Loh R S K, MacLennan S. et al Comparison of autologous serum eye drops with conventional therapy in a randomised controlled crossover trial for ocular surface disease. Br J Ophthalmol 2004. 88647–652.652 [PMC free article] [PubMed]
22. Wright P. Topical retinoic acid therapy for disorders of the outer eye. Trans Ophthalmol Soc UK 1985. 104869–874.874 [PubMed]
23. Soong H Z, Martin N F, Wagoner M D. et al Topical retinoid therapy for squamous metaplasia of various ocular surface disorders. Ophthalmology 1988. 951442–1446.1446 [PubMed]
24. Kawasaki S, Nishida K, Sotozono Ch. et al Conjunctival inflammation in the chronic phase of Stevens–Johnson syndrome. Br J Ophthalmol 2000. 841191–1193.1193 [PMC free article] [PubMed]
25. Romero‐Rangel T, Stavrou P, Cotter J. et al Gas‐permeable scleral contact lens therapy in ocular surface disease. Am J Ophthalmology 2000. 13025–32.32
26. Tappin M J, Pullum K W, Buckley R J. Scleral contact lenses for overnight wear in the management of ocular surface disorders. Eye 2001. 15168–172.172 [PubMed]
27. Ormerod L D, Fong L P, Foster C S. Corneal infection in mucosal scarring disorders and Sjogren's syndrome. Am J Ophthalmol 1988. 105512–518.518 [PubMed]
28. Shimazaki J, Shimmura S, Fujishima H. et al Association of preoperative tear function with surgical outcome in severe Stevens–Johnson syndrome. Ophthalmology 2000. 1071518–1523.1523 [PubMed]
29. Gomes J A P, Santos M S, Ventura A S. et al Amniotic membrane with living related corneal limbal/conjunctival allograft for ocular surface reconstruction in Stevens–Johnson syndrome. Arch Ophthalmol 2003. 1211369–1374.1374 [PubMed]
30. Samson C M, Nduaguba C, Baltatzis S. et al Limbal stem cell transplantation in chronic inflammatory eye disease. Ophthalmology 2002. 109862–868.868 [PubMed]
31. Tugal‐Tutkun I, Aydin Akova Y, Foster A S. Penetrating keratoplasty in cicatrising conjunctival diseases. Ophthalmology 1995. 102576–585.585 [PubMed]
32. Yaghouti F, Nouri M, Abad J C. et al Keratoprosthesis: preoperative prognostic categories. Cornea 2001. 2019–23.23 [PubMed]
33. Geerling G, Sieg P, Bastian G O. et al Transplantation of the autologous submandibular gland for most severe cases of keratoconjunctivitis sicca. Ophthalmology 1998. 105327–335.335 [PubMed]
34. Geerling G, Liu C S, Dart J K. et al Sight and comfort: complex procedures in end‐stage Stevens‐Johnson syndrome. Eye 2003. 1789–91.91 [PubMed]

Articles from The British Journal of Ophthalmology are provided here courtesy of BMJ Publishing Group