Glucocorticoids (GCs) are often used to treat Rheumatoid Arthritis (RA) despite their many side effects and the availability of other effective therapies. Cataract and glaucoma are known side effects of GCs but the risk of them developing in the setting of GC use for RA is unknown. The aim was to perform a systematic review and meta-analysis to determine the association between GCs and the risk of developing cataract and/or glaucoma in RA.
A systematic search was carried out using MEDLINE, EMBASE, and Web of Science. All RCTs comparing GC use to non-use in RA populations were sought. Observational studies reporting cataract and/or glaucoma amongst GC users and non-users were also included. Data extracted included incidence/prevalence of cataract and/or glaucoma in each arm, dose and duration of therapy. Two independent reviewers performed quality assessment.
28 RCTs met eligibility criteria, however only 3 reported cataracts and glaucoma, suggesting significant under-reporting. An association between GC use and the development of cataracts in RA patients was seen in observational studies but not RCTs. There was no statistically significant association between GC use and the development of glaucoma, although data were sparse. There were insufficient data to determine the impact of dose and duration of therapy.
The current literature suggests a possible association between GC use and the development of cataract. However, this risk cannot be accurately quantified in RA from the available evidence. RCTs have not adequately captured these outcomes and well-designed observational research is required.
Citation: Black RJ, Hill CL, Lester S, Dixon WG (2016) The Association between Systemic Glucocorticoid Use and the Risk of Cataract and Glaucoma in Patients with Rheumatoid Arthritis: A Systematic Review and Meta-Analysis. PLoS ONE 11(11): e0166468. https://doi.org/10.1371/journal.pone.0166468
Editor: Graham R. Wallace, University of Birmingham, UNITED KINGDOM
Received: May 23, 2016; Accepted: October 28, 2016; Published: November 15, 2016
Copyright: © 2016 Black et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: RJB was the recipient of the 2013 Ken Muirden Overseas Training Fellowship, an educational research grant jointly funded by Arthritis Australia and Roche Products Pty Ltd. RJB was also supported by an Australian Postgraduate Award (The University of Adelaide) and an Arthritis South Australia/Australian Rheumatology Association (South Australian Branch) Postgraduate Grant. WGD was supported by a MRC Clinician Scientist Fellowship (G0902272). We thank the Arthritis Research UK for their support: Arthritis Research UK grant number 20380. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: RJB was the recipient of the 2013 Ken Muirden Overseas Training Fellowship, an educational research grant jointly funded by Arthritis Australia and Roche Products Pty Ltd. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Cataract and glaucoma were first described as side effects of systemic GC therapy as early as 1953[1, 2]. In particular, posterior subcapsular cataracts (PSCs) are a subtype of cataracts that occur more frequently in GC-exposed patients . Similarly, GC exposure can lead to steroid-induced glaucoma, a type of open angle glaucoma. These conditions can lead to visual impairment, resulting in significant disability and cost to the healthcare system . As GCs remain widely prescribed in rheumatoid arthritis (RA) and many other inflammatory diseases [4, 5], any increased risk might lead to a significant public health burden.
EULAR guidelines advocate that clinicians inform their patients of the risk of side effects associated with GCs before commencing treatment . However, for cataract and glaucoma, the magnitude of risk is rarely reported and current literature has not been reviewed to determine if the risk can be accurately quantified. Specific questions, such as how this is influenced by dose and duration of therapy, also have not yet been addressed. RA was the first condition to be treated with GCs in 1948  and it was also the first condition in which PSCs were described as AEs of GC use in 1950 . There are multiple randomised controlled trials (RCTs) of GC use in RA, and long-term use has been looked at in many observational studies, making it an ideal setting to explore these questions.
The aim was to perform a systematic literature review and meta-analysis of RCTs and observational studies examining the association between GC use and the risk of cataract and glaucoma in patients with RA compared to patients with RA not exposed to GCs. Secondary aims were to determine whether there is an association with dose and duration of GC therapy.
Materials and Methods
A systematic search was conducted using MEDLINE, EMBASE and Web of Science for articles published to January 2016. There was no review protocol for this systematic review. Separate search strategies were conducted for RCTs and observational studies. RCTs fulfilling the following criteria were included: 1) RA population 2) Exposure to systemic (oral, intramuscular or intravenous) GC therapy in one arm and non-exposure (no treatment or placebo) in at least one comparator arm. Studies comparing DMARD(s) plus GC in one arm and the same DMARD(s) without GC in the comparison arm were also included. Inclusion criteria for observational studies were: 1) RA population, 2) Use of a cross-sectional, case-control or cohort study design 3) Reporting of the number or rate of cataracts and/or glaucoma in GC-exposed and non-exposed patients. Although cross-sectional studies do not provide information on causality, they were included in this review in order to capture information from about PSCs from early studies conducted in the 1960s. Studies conducted exclusively in populations other than RA (including early inflammatory arthritis, undifferentiated polyarthritis and JIA) were excluded. If RA was one of several indications reported, the study was only included if the RA population was reported separately from the other populations. Exposure was restricted to systemic GC therapy (oral, intramuscular, intravenous), and studies reporting only topical, intra-articular, intra-ocular or other non-systemic routes of steroid administration were excluded. In instances where two studies reported on the same cohort, the earlier of the two studies was retained unless the latter reported on cataracts and/or glaucoma.
Search terms are listed in the online supporting information (S1 Table). Search filters for RCTs were based on the Cochrane Highly Sensitive Search Strategy for identifying randomized trials in MEDLINE: sensitivity-maximizing version .The search filters for observational studies was based on the Scottish Intercollegiate Guidelines Network search filters for observational studies .
Study Selection and Data extraction
The initial study selection was based on review of title and abstract. Studies in non-RA populations were excluded at this stage, as were those with designs other than RCTs, cohort, case control or cross-sectional studies. Only articles published in English were selected due to lack of access to a translation service. Full text manuscripts of all remaining articles were reviewed for eligibility based on the inclusion and exclusion criteria. Hand searching of references of all papers obtained through the above search and relevant review articles was also carried out. Abstract only publications and unpublished studies were not considered. A random sample of 20 RCTs was selected for independent full-text review by the second reviewer (CH). Each study was assessed in regards to eligibility criteria and the rate of agreement with the first reviewer (RB) was then determined.
Data extraction was performed by a single reviewer using an electronic data collection form. Information was extracted regarding: 1. The RA population including age, gender and disease duration, 2. The exposure including the definition of GC use, dose and duration of therapy, 3. The comparator arm including whether this was placebo or non-GC exposure, and 4. The outcome of cataract and glaucoma in each arm, how this was obtained and whether this was reported as point prevalence, prevalence, incidence or incidence rate.
Both reviewers (RB, CH) performed an independent assessment of the quality of all included studies. Discrepancies were discussed and resolved by consensus. The Cochrane Risk of Bias Assessment tool  was used to assess the quality of all included RCTs reporting cataract and/or glaucoma. Cohort studies were assessed using the Newcastle Ottawa Scale and each of the cross-sectional studies was assessed for selection bias, detection bias, attrition bias, and reporting bias, based on the Cochrane Handbook common classification scheme for bias. No case control studies meeting eligibility criteria were identified in the search. The RCTs and cross-sectional studies were assessed as low, high or uncertain risk, however the cohort studies were assessed as either being low or high risk due to differences in the scales used.
Although data were limited, meta-analyses were performed as a means of summarising the data available. Cross sectional studies were not considered in the meta-analysis due to their inability to assess causality. Random effects meta-analyses were performed for each outcome: all cataracts and glaucoma. While other methodologies for dealing with zero events and sparse data were considered [13–15], risk difference (RD) was selected as the effect size because many studies had zero events in one arm, and odds ratios and relative risks are not defined in this setting. Statistical heterogeneity was assessed using the I2 statistic, where I2>50% represents significant heterogeneity. Meta analyses were carried out using the Metafor package in R .
Effect of dose and duration of GC exposure and visual outcomes
A description, rather than formals analysis of the effect of dose and duration of GC use on the development of cataract and glaucoma was carried out due to the scarcity of studies addressing this.
Comparing the reporting of cataracts in RCTs, to population-based incidence of cataracts
Concerns that cataracts were not being captured or reported in RCTs, early in the course of this review, led to a secondary aim to compare the RCT incidence with the expected incidence of cataracts based on general population rates derived from The Blue Mountains Eye Study (BMES) . For comparison with RA RCTs, the expected population cumulative incidence of cataracts over 1 and 2 years was estimated for patients aged less than 75 years from the reported BMES 5-year incidence, under the constant hazards assumption.
Randomised Controlled Trials
95 RCTs were identified for full-text review and 28 RCTs were ultimately included after applying eligibility criteria (Fig 1). There was initial agreement for 19/20 articles selected for assessment by the second reviewer. Disagreement occurred for one article because the results of this study had already been reported earlier in another paper that had not been included in the random sample. Once this issue was made known to the second reviewer, there was full consensus for all articles. These papers are summarised in the supporting information (S2 Table).
Flow charts depicting selection of A. Randomised Controlled Trials and B. Observational Studies.
Cataracts and glaucoma were reported in only 3 of the 28 RCTs (Table 1), each of 2 years’ duration. The initial 28 RCTs evaluated treatment with oral, intramuscular and intravenous GCs, however the final three studies all assessed oral exposure. These three studies reported all cataracts rather than the PSC subtype. One study, Williams et al , was excluded because it did not report whether the single observed glaucoma case occurred in the GC or control group. The remaining 24 RCTs did not mention cataracts and/or glaucoma in either the methods or the results. The protocols for all three RCTs reporting incident eye disease included either recording specific known GC-associated AEs using a standardised list [19, 20] or ophthalmological examinations [20, 21]. There were no significant differences in the risk of developing cataracts for GC-exposed versus unexposed RA patients, either individually by trial, or collectively in the meta-analysis (RD 0.01 events/patient, 95% CI -0.01–0.03), Fig 2 (A RD of 0.01 equates to an additional 10 cataracts for every 1000 patients exposed to GCs compared to those unexposed). Similarly, there were no significant differences in the risk of developing glaucoma for GC-exposed versus unexposed RA patients, either individually by trial, or collectively in the meta-analysis (RD 0.01 events/patient, 95% CI -0.02–0.04), Fig 2.
Comparing the reporting of cataracts in RCTs to population-based incidence of cataracts
Of the 28 RCTS comparing GC use to non-use in RA, there were a total of 13 cataracts reported in only 3/28 studies (11%). The remaining 25 studies did not report cataracts in their results, nor did they set out to do so in the their methods. Across all 28 RCTs there was a total of 4604 person years, giving a combined cataract incidence rate of 2.8 per 1000 person years (pyr).
Considering incidence only in the three studies that reported cataracts, the random effects estimate of the combined incidence rate (IR) was 11/1000pyr (95%CI 3-43/1000pyr), although the IRs ranged considerably from 2.1/1000pyr  to 12.3/1000pyr  to 26.5/1000pyr . This heterogeneity is likely related to the different methods used to detect cataract, with ophthalmological examination detecting a greater prevalence of cataracts.
The cumulative two-year incidence in the 3 studies of this duration reporting cataract was 2.0% (95% CI 0.6–6.6). The expected population cumulative incidence (in participants younger than 75 years), estimated from the BMES population , is 6% over 1 year and 12% over two years. Given that the median duration of the 27 RCTs was one year, this suggests that cataracts were expected in RCTS in which they were not reported, suggesting a substantial six-fold under-reporting of cataracts in the clinical trials.
A total of 10 observational studies met eligibility criteria for the final review. Five were cohort studies, of which two were prospective and three retrospective. The remaining five were cross-sectional (Fig 1).
a) All Cataracts (Reported in 5 cohort studies).
Amongst the cohort studies, Gullberg et al  reported PSCs as well as all cataracts. However, the remaining cohort studies reported all cataracts only and not PSCs [23–26]. The methods for identifying and defining cataracts varied between studies, as did the definitions of GC use (Table 2).
In contrast to RCTs, four of the five cohort studies reported a significantly increased risk of cataracts in GC-exposed patients, and collectively the combined risk difference was 0.07 events per person (95%CI 0.04–0.10), Fig 2, which is equivalent to an additional 70 cataracts seen per 1000 patients in those exposed to GCs. The odds ratio, which could also be meaningfully estimated from this data, was 2.1 (95%CI 1.5–2.9). While there was considerable variation in age, disease duration, dose and duration on GCs between these studies (Table 2), there was surprisingly little statistical evidence of heterogeneity between the effect sizes (RD I2 = 6.2%).
In regards to dose and duration of therapy, Huscher et al  looked at patterns of glucocorticoid induced side effects and found two distinct patterns including ‘linear risk’, where risk of developing an AE increased with dose in a linear fashion and ‘threshold risk’ where the increased risk only occurred after a threshold dose was reached. They found that both cataract and glaucoma fit the threshold pattern with a threshold dose of 5mg per day for cataract and 7.5mg per day for glaucoma in patients on GCs for more than 6 months. Mazzantini et al  did not find any difference in cataract prevalence between those treated with GCs for <2 years, 2–5 years or >5years.
b) Posterior Supcapsular Cataracts (PSCs) (Reported in 5 cross-sectional studies and 1 cohort study).
PSCs were reported in five cross-sectional studies published between 1960 and 1969. In all five studies, an RA population, including those exposed and unexposed to GC treatment, underwent ophthalmological examination. In 4/5 of these studies[1, 27–29], PSCs were only seen in those exposed to GCs, and in the remaining study by Williamson et al , 1 PSC was seen amongst 159 unexposed to GCs compared to 10/148 exposed (Table 3). PSCs were also reported in a retrospective cohort study by Gullberg et al , with a risk difference of 0.20 events per patient (95%CI 0.08–0.33), equivalent to an additional 200 PSCs for every 1000 patients exposed to GCs.
In regards to dose and duration of GC therapy, Black et al reported that PSCs were not seen in six patients who took prednisolone equivalent doses of <10mg per day or in nine patients who had been taking GCs for less than one year . Giles et al  also found that PSCs did not occur in three patients that received GCS for less than one year. However they did occur in four of the 19 patients taking <10mg per day for ≥6 months. The prevalence of PSCs increased with GC dose, with PSCs seen in 4/12 patients on 10-15mg/day and in 6/7 patients on >15mg per day. Furst et al compared the prednisolone equivalent average daily dose and duration of therapy in those with PSCs to those without and found no significant difference in duration of therapy but a higher average dose of 11.5mg in those that developed PSCs compared to 8.8mg in those that did not. However, they did not report whether this difference was significant. Williamson also reported an increased prevalence with longer duration of therapy, with no PSCs seen in the 65 patients on GCs for less that two years compared to seven of the 52 patients on GCs for more than two years (p<0.01) .
c) Glaucoma (Reported in2 cohort studies and 1 cross-sectional study).
Of the observational studies that fulfilled the eligibility criteria for this search, three studies reported glaucoma [23, 25, 30]. No significant association was seen between GC use and glaucoma amongst the 2 cohort studies [23, 25] with a combined risk difference of 0.00 (95% CI -0.01–0.02). In the cross-sectional study, Williamson et al  reported 1 case of glaucoma amongst 148 GC users and 1 case amongst 159 non-users.
The quality assessment of all studies is shown in Fig 3. Among RCTs, the risk of bias was low for most criteria (Fig 3A). Conversely, for the cohort studies, there was a high risk of bias in some categories, particularly regarding the exclusion of prevalent eye disease (Fig 3B). Amongst the cross-sectional studies, the risk of bias was uncertain for most categories (Fig 3C).
Quality Assessment of A. Randomised controlled trials and C. Cross sectional studies using the Cochrane Risk of Bias Assessment Tool (high risk, low risk of uncertain risk) and B. Cohort studies using the Newcastle Ottawa Scale (high risk or low risk).
RA is an ideal patient group in which to look for long term adverse effects of GCs. RA was the first condition to be treated with GCs and it was also the first condition in which PSCs were described as AEs of GC use . More recently GCs have been shown to have a disease modifying effect in RA , resulting in more continuous patterns of use compared to other indications such as asthma and IBD for which GCs are often used intermittently. RA patients may differ in their risk of developing GC induced eye diseases compared to other populations such as those with IBD and spondyloarthropathies, which are more frequently associated with uveitis and topical GC use, other known risk factors for developing cataract and glaucoma . It is therefore important to understand how systemic GC use affects the risk of developing cataract and glaucoma in RA patients.
This review has found that there is limited literature addressing the association between systemic glucocorticoid use and the risk of cataract and glaucoma in patients with RA. In contrast to the RCTs which showed no association between GC use and all cataracts (RD 0.01, 95%CI -0.01–0.03), an association was observed in the observational cohort studies [22–26], with a estimated risk difference of 0.07, 95%CI 0.04–0.10. The RCT and observational literature regarding glaucoma was particularly limited as was that addressing the effect of dose and duration of GC therapy. There were no studies addressing the impact of comorbidities known to be associated with increased risk of cataract or glaucoma, or the visual outcome of GC-induced cataract or glaucoma. PSCs often require surgery at an earlier stage than other cataract types  and whilst this has not been addressed in the context of GC use, it certainly raises concern and highlights the need for future studies in this area.
Although well described as side effects of GC use in review articles and books [34–38], cataracts have been reported in only three of the 28 RCTs comparing GC use and non-use in RA populations (in either arm of the study). Through comparison to general population cumulative incidence rates derived from the BMES, it is apparent there is significant under-reporting of cataracts in RA glucocorticoid clinical trials, making any estimates of GC-associated risk less reliable. Ideally, future RCTs comparing GC use to non-use would endeavour to accurately capture this information with regular ophthalmology assessments.
Whilst this review focussed specifically on studies in RA populations, it is interesting to consider how the extent of GC-associated risk compares with studies in other populations, and between the sub-types of cataract. In the BMES Australian general population there was an association between current GC use and PSCs (OR 4.11, 95%CI 1.67–10.08) and nuclear cataracts (OR 3.45, 95% CI 1.26–9.43) but not cortical cataracts (OR 1.08, 95%CI 0.44–2.64) . In comparison, the Beaver Dam Eye Study found an increased incidence of cortical cataracts associated with GC use in the American general population (OR 2.59, 95%CI 1.45–4.62) but not PSCs (OR 1.27, 95%CI 0.42–3.86) or nuclear cataracts (OR 1.41, 95%CI 0.77–2.56) . This is in contrast to other studies that have shown a strong association between GC use and PSCs in other populations including asthma and renal transplantation [33, 40–43]. Unfortunately, despite the observation that PSCs occurred with GC use reported in earlier cross-sectional studies, more recent RA studies have not reported specific cataract type.
Impact of Dose and Duration of GC Therapy
The impact of dose and duration of GC therapy on the development of cataract and glaucoma remains unanswered. Whilst some of the observational studies in this review suggest an association between GC use and dose and duration of therapy, they do not adequately quantify the risk in a way that can be translated into clinical practice when discussing the risks and benefits of GC treatment for RA. Many of the observational studies that comment on dose and duration were insufficiently powered, or not appropriately designed, to determine an accurate association between different doses and durations of therapy. A striking association between GC use and the development of PSCs was described in early cross-sectional studies carried out in the 1960s, however an association with cataracts was not seen in more recent RCTs. Whilst under-reporting partly accounts for this, another possible explanation is that GC exposure (dose and duration of therapy) has declined over time with the increasing use of concurrent DMARDs, highlighting the need for this to be explored in future studies.
Strengths and limitations
The main strength of this review is the systematic approach used to identify all relevant studies published since 1950 and the use of two reviewers to confirm eligibility criteria. The main limitation of this review is the small number of studies meeting eligibility criteria, making it impossible to draw meaningful conclusions. Although cross-sectional studies cannot assess causality, their inclusion allowed a comprehensive summary of the complete literature on this subject. The results are reported by study design and thus this does not detract from the study findings. The heterogeneity of studies was significant, particularly in regards to the methods used to detect cataracts and glaucoma. Only one of the RCTs and two of the cohort studies used examinations to detect cataracts, and the remainder relied on patient reports or case note review. There was additional heterogeneity in the definitions of GC exposure. This heterogeneity made it impossible to draw any quantitative conclusions regarding the risk of GC use and the development of cataract and glaucoma in RA patients. The observational studies had a high risk of bias, with only one  controlling for comorbidities known to be associated the development of cataract and glaucoma and potentially associated with GC exposure [44–46].
The current literature leaves many questions unanswered in regards to GC use and the development of cataract and glaucoma. Although RCTs often represent the highest level of evidence for efficacy, they are not ideal for examining long-term drug adverse effects. They also do not reflect the patterns of GC use seen in routine practice. In order to adequately address the questions in this review, a well designed observational study that is large enough and of sufficient duration to capture rare events is required. Ideally, such a study would be conducted in a real-life setting so that the impact of age and comorbidities could also be determined. Data on dose and duration of GC use would need to be accurately captured and appropriate methodologies employed to fully understand the interactions between dose, duration and recentness of treatment. Cataracts and glaucoma become increasingly prevalent with age and commonly occur without exposure to systemic GCs. They range in severity from asymptomatic to visually disabling, making it important to clearly define the clinical relevance of cataract/glaucoma in future studies.
In conclusion, observational studies, but not RCTs, suggest an association between GC use and the development of cataracts in RA patients. This relationship has not been adequately quantified, nor have the effect of dose and duration of therapy, visual outcomes and the impact of comorbidities been addressed. Confidence in RCT and observational estimates are limited by under-reporting, and heterogeneity and bias, respectively. There is no robust literature regarding GC use and the development of glaucoma in RA patients. These findings highlight the need for future well conducted observational studies targeting the unanswered questions as well as improved capture of these outcomes in future RCTs.
S1 Table. Search terms used for RCTs and Observational Studies in each of the databases.
S2 Table. RCTs comparing GC use to non-use in RA and number of reported cataract and glaucoma.
- Conceptualization: RJB WGD.
- Formal analysis: RJB SL.
- Investigation: RJB CLH.
- Methodology: RJB CLH SL WGD.
- Supervision: CLH WGD.
- Visualization: RJB SL.
- Writing – original draft: RJB.
- Writing – review & editing: RJB CLH SL WGD.
- 1. Black RL, Oglesby RB, Von Sallmann L, Bunim JJ. Posterior subcapsular cataracts induced by corticosteroids in patients with rheumatoid arthritis. JAMA. 1960;174:166–71. Epub 1960/09/10. pmid:13801171.
- 2. Stern JJ. Acute glaucoma during cortisone therapy. American journal of ophthalmology. 1953;36(3):389–90. pmid:13030647.
- 3. Taylor HR, Keeffe JE, Vu HT, Wang JJ, Rochtchina E, Pezzullo ML, et al. Vision loss in Australia. The Medical journal of Australia. 2005;182(11):565–8. Epub 2005/06/09. pmid:15938683.
- 4. Black RJ, Joseph RM, Brown B, Movahedi M, Lunt M, Dixon WG. Half of UK patients with rheumatoid arthritis are prescribed oral glucocorticoid therapy in primary care: a retrospective drug utilisation study. Arthritis research & therapy. 2015;17(1):375. pmid:26702817; PubMed Central PMCID: PMC4718024.
- 5. Fardet L, Petersen I, Nazareth I. Prevalence of long-term oral glucocorticoid prescriptions in the UK over the past 20 years. Rheumatology (Oxford). 2011;50(11):1982–90. pmid:21393338.
- 6. Hoes JN, Jacobs JW, Boers M, Boumpas D, Buttgereit F, Caeyers N, et al. EULAR evidence-based recommendations on the management of systemic glucocorticoid therapy in rheumatic diseases. Annals of the rheumatic diseases. 2007;66(12):1560–7. Epub 2007/07/31. pmid:17660219; PubMed Central PMCID: PMC2095301.
- 7. Hench PS, Kendall EC, Slocumb CH, Polley HF. Effects of Cortisone Acetate and Pituitary Acth on Rheumatoid Arthritis, Rheumatic Fever and Certain Other Conditions—A Study in Clinical Physiology. Archives of internal medicine. 1950;85(4):545–666. pmid:15411248
- 8. The Cochrane highly sensitive search strategies for identifying randomized trials in PubMed [cited 2015 October]. Available from: http://work.cochrane.org/pubmed.
- 9. Scottish Intercollegiate Guidelines Network Search Filters [cited 2015 October]. Available from: http://www.sign.ac.uk/methodology/filters.html.
- 10. Table 8.5.a: The Cochrane Collaboration's tool for assessing risk of bias [cited 2015 October]. Available from: http://www.handbook.cochrane.org.
- 11. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses [cited 2015 October]. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
- 12. Table 8.4.a: A common classification scheme for bias. [cited 2015 Oct]. Available from: http://www.handbook.cochrane.org.
- 13. Stijnen T, Hamza TH, Ozdemir P. Random effects meta-analysis of event outcome in the framework of the generalized linear mixed model with applications in sparse data. Statistics in medicine. 2010;29(29):3046–67. pmid:20827667.
- 14. Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? Use and avoidance of continuity corrections in meta-analysis of sparse data. Statistics in medicine. 2004;23(9):1351–75. pmid:15116347.
- 15. Bradburn MJ, Deeks JJ, Berlin JA, Russell Localio A. Much ado about nothing: a comparison of the performance of meta-analytical methods with rare events. Statistics in medicine. 2007;26(1):53–77. pmid:16596572.
- 16. R Core Team (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. [cited 2016 July]. Available from: http://www.R-project.org/.
- 17. Kanthan GL, Wang JJ, Rochtchina E, Tan AG, Lee A, Chia EM, et al. Ten-year incidence of age-related cataract and cataract surgery in an older Australian population. The Blue Mountains Eye Study. Ophthalmology. 2008;115(5):808–14 e1. Epub 2007/09/29. pmid:17900695.
- 18. Williams IA, Baylis EM, Shipley ME. A double-blind placebo-controlled trial of methylprednisolone pulse therapy in active rheumatoid disease. Lancet. 1982;2(8292):237–40. Epub 1982/07/31. pmid:6124671.
- 19. Bakker MF, Jacobs JWG, Welsing PMJ, Verstappen SMM, Tekstra J, Ton E, et al. Low-dose prednisone inclusion in a methotrexate-based, tight control strategy for early rheumatoid arthritis: a randomized trial. Annals of internal medicine. 2012;156(5):329–39. pmid:22393128.
- 20. van Everdingen AA, Jacobs JWG, Siewertsz Van Reesema DR, Bijlsma JWJ. Low-dose prednisone therapy for patients with early active rheumatoid arthritis: clinical efficacy, disease-modifying properties, and side effects: a randomized, double-blind, placebo-controlled clinical trial. Annals of internal medicine. 2002;136(1):1–12. pmid:11777359.
- 21. Wassenberg S, Rau R, Steinfeld P, Zeidler H. Very low-dose prednisolone in early rheumatoid arthritis retards radiographic progression over two years: a multicenter, double-blind, placebo-controlled trial. Arthritis and rheumatism. 2005;52(11):3371–80. pmid:16255011.
- 22. Gullberg R, Elman A. Posterior subcapsular cataract in rheumatoid arthritis patients after long-term, low-dosage paramethasone therapy. Scand J Rheumatol. 1973;2(3):113–8. pmid:4769063.
- 23. Saag KG, Koehnke R, Caldwell JR, Brasington R, Burmeister LF, Zimmerman B, et al. Low dose long-term corticosteroid therapy in rheumatoid arthritis: an analysis of serious adverse events. The American journal of medicine. 1994;96(2):115–23. Epub 1994/02/01. pmid:8109596.
- 24. McDougall R, Sibley J, Haga M, Russell A. Outcome in patients with rheumatoid arthritis receiving prednisone compared to matched controls. The Journal of rheumatology. 1994;21(7):1207–13. pmid:7966059.
- 25. Huscher D, Thiele K, Gromnica-Ihle E, Hein G, Demary W, Dreher R, et al. Dose-related patterns of glucocorticoid-induced side effects. Annals of the rheumatic diseases. 2009;68(7):1119–24. Epub 2008/08/08. pmid:18684744.
- 26. Mazzantini M, Talarico R, Doveri M, Consensi A, Cazzato M, Bazzichi L, et al. Incident Comorbidity Among Patients with Rheumatoid Arthritis Treated or Not with Low-dose Glucocorticoids: A Retrospective Study. The Journal of rheumatology. 2010;37(11):2232–6. pmid:WOS:000284301900007.
- 27. Giles CL, Mason GL, Duff IF, McLean JA. The association of cataract formation and systemic corticosteroid therapy. JAMA. 1962;182:719–22. pmid:13947854.
- 28. Crews SJ. Posterior subcapsular lens opacities in patients on longterm corticosteroid therapy. British medical journal. 1963;1(5346):1644–7. Epub 1963/06/22. pmid:14023840; PubMed Central PMCID: PMC2124356.
- 29. Furst C, Smiley WK, Ansell BM. Steroid cataract. Annals of the rheumatic diseases. 1966;25(4):364–8. Epub 1966/07/01. pmid:5947583; PubMed Central PMCID: PMC2453341.
- 30. Williamson J, Paterson RW, McGavin DD, Jasani MK, Boyle JA, Doig WM. Posterior subcapsular cataracts and glaucoma associated with long-term oral corticosteroid therapy. In patients with rheumatoid arthritis and related conditions. The British journal of ophthalmology. 1969;53(6):361–72. Epub 1969/06/01. pmid:5794952; PubMed Central PMCID: PMC1214577.
- 31. Kirwan JR, Bijlsma JW, Boers M, Shea BJ. Effects of glucocorticoids on radiological progression in rheumatoid arthritis. Cochrane database of systematic reviews. 2007;(1):CD006356. Epub 2007/01/27. pmid:17253590.
- 32. Rothova A, Suttorp-van Schulten MS, Frits Treffers W, Kijlstra A. Causes and frequency of blindness in patients with intraocular inflammatory disease. The British journal of ophthalmology. 1996;80(4):332–6. pmid:8703885; PubMed Central PMCID: PMC505460.
- 33. Wang JJ, Rochtchina E, Tan AG, Cumming RG, Leeder SR, Mitchell P. Use of inhaled and oral corticosteroids and the long-term risk of cataract. Ophthalmology. 2009;116(4):652–7. Epub 2009/02/27. pmid:19243828.
- 34. Hodge WG, Whitcher JP, Satariano W. Risk factors for age-related cataracts. Epidemiologic reviews. 1995;17(2):336–46. Epub 1995/01/01. pmid:8654515.
- 35. Caldwell JR, Furst DE. The efficacy and safety of low-dose corticosteroids for rheumatoid arthritis. Seminars in arthritis and rheumatism. 1991;21(1):1–11. Epub 1991/08/01. pmid:1948096.
- 36. Carnahan MC, Goldstein DA. Ocular complications of topical, peri-ocular, and systemic corticosteroids. Current opinion in ophthalmology. 2000;11(6):478–83. Epub 2001/01/06. pmid:11141645.
- 37. Da Silva JA, Jacobs JW, Kirwan JR, Boers M, Saag KG, Ines LB, et al. Safety of low dose glucocorticoid treatment in rheumatoid arthritis: published evidence and prospective trial data. Annals of the rheumatic diseases. 2006;65(3):285–93. Epub 2005/08/19. pmid:16107513; PubMed Central PMCID: PMC1798053.
- 38. Jacobs JWG, Bijlsma JWJ. Glucocorticoid Therapy. In: Firestein GS, Budd RC, Gabriel S, McInnes I, O'Dell JR, editors. Kelley's Textbook of Rheumatology. 1. 9th Edition ed. Philadelphia: Elsevier Saunders; 2013. p. 894–916.
- 39. Klein BE, Klein R, Lee KE, Danforth LG. Drug use and five-year incidence of age-related cataracts: The Beaver Dam Eye Study. Ophthalmology. 2001;108(9):1670–4. pmid:11535471.
- 40. Leske MC, Chylack LT Jr., Wu SY. The Lens Opacities Case-Control Study. Risk factors for cataract. Arch Ophthalmol. 1991;109(2):244–51. pmid:1993036.
- 41. Group TI-ACS. Risk factors for age-related cortical, nuclear, and posterior subcapsular cataracts. American journal of epidemiology. 1991;133(6):541–53. pmid:1672483.
- 42. Toogood JH, Markov AE, Baskerville J, Dyson C. Association of ocular cataracts with inhaled and oral steroid therapy during long-term treatment of asthma. The Journal of allergy and clinical immunology. 1993;91(2):571–9. pmid:8436773.
- 43. Matsunami C, Hilton AF, Dyer JA, Rumbach OW, Hardie IR. Ocular complications in renal transplant patients. Aust N Z J Ophthalmol. 1994;22(1):53–7. pmid:8037915.
- 44. Delcourt C, Cristol JP, Tessier F, Leger CL, Michel F, Papoz L. Risk factors for cortical, nuclear, and posterior subcapsular cataracts: the POLA study. Pathologies Oculaires Liees a l'Age. American journal of epidemiology. 2000;151(5):497–504. pmid:10707918.
- 45. Leske MC, Connell AM, Wu SY, Hyman LG, Schachat AP. Risk factors for open-angle glaucoma. The Barbados Eye Study. Arch Ophthalmol. 1995;113(7):918–24. pmid:7605285.
- 46. Kersey JP, Broadway DC. Corticosteroid-induced glaucoma: a review of the literature. Eye. 2006;20(4):407–16. Epub 2005/05/07. pmid:15877093.