Inhaled Corticosteroid Exposure and Risk of Cataract in Patients with Asthma and COPD: A Systematic Review and Meta-Analysis

Purpose Both systemic and inhaled corticosteroids may increase the risk of cataract in patients with both chronic obstructive pulmonary disease (COPD) and asthma. Our aim was to assess the degree of association between cataract and corticosteroid exposure in patients with asthma and COPD. Methods A systematic literature review and meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines. The odds ratio estimates were extracted from each article. A random effects model was applied for estimate pooling in separate meta-analyses according to study design. Meta-regression was performed to assess the dose-response relationship between corticosteroid exposure and the risk of cataract development. Results A total of 19 studies met the criteria for inclusion in this review, of which 12 studies provided effect estimates for pooled analyses. All but one of the included observational studies reported a significant association between use of corticosteroids and cataract development in cohorts of asthma and/or COPD patients. Pooled analyses revealed on average a doubled risk of cataract in corticosteroid-exposed asthma and COPD patients. Studies have shown that daily high-dose inhaled corticosteroid (ICS) ≥ 1000 μg is associated with a significant risk of developing cataract and by that predispose to subsequent cataract surgery, although one study showed that systemic corticosteroids increase cataract risk more than ICS. Conclusion ICS treatment in asthma and COPD patients is a risk factor for cataract development. Our results emphasize a previously underestimated potential long-term risk of treatment with ICS and underline the importance of targeting ICS treatment, and not least dosing, to improve the risk-benefit ratio of maintenance treatment in both asthma and COPD.


Introduction
Cataract is a prevalent eye condition characterized by an increasingly opaque lens over time leading to blurred vision [1,2].Cataract is the most common treatable cause of acquired blindness worldwide [3].Eye examination showing the presence of cataract [4] and surgery is the treatment of choice when loss of vision afects daily activities [5,6].
It has for decades been reported that long-term use of systemic corticosteroids may be associated with the development of cataract.Te frst study was published almost six decades ago [7].Te type of cataract usually associated with long-term systemic corticosteroid treatment is posterior subcapsular cataract (PSC) [8,9].
A possible association between cataract development and long-term treatment with inhaled corticosteroids (ICS) for asthma and chronic obstructive pulmonary disease (COPD) has been attracting considerable interest [10][11][12][13][14]. Te association has been assessed in a recent review [15].However, since then, more studies have been published addressing the possible association between cataract development and corticosteroid treatment in asthma and COPD [11,12,14,[16][17][18].
A challenging aspect of both obstructive airways diseases and cataract is that they both cause great morbidity and afect millions of people worldwide [19,20].Te risk of developing cataract due to ICS treatment for asthma and COPD is most likely lower than cataract risk due to systemic corticosteroid treatment, but this has only been addressed in very few studies [21,22].
Several studies have proposed a high prevalence of comorbidities among cataract patients, including respiratory diseases such as asthma and COPD [10,[23][24][25][26]. Our aim was to explore the possible association of ICS therapy in asthma and COPD and the risk of cataract development.In order to reduce the risk of development of cataract in asthma and COPD patients, it is of great importance for both patients with obstructive airways disease and society to have access to the bestpossible knowledge of the risk factors for cataract development.
In this systematic review and meta-analysis, we provide an update on current knowledge regarding the cataract risk in COPD and asthma patients associated with their prescribed systemic and/or inhaled corticosteroids.

Materials and Methods
2.1.Search Strategy.Tis systematic review was performed and reported in agreement with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [27].
Studies were included if they fulflled the following inclusion criteria: (i) Original research reporting on cataract risk in asthma and/or COPD patients prescribed systemic and/or inhaled corticosteroids (ii) Studies published within the last three decades Studies were not included if they fulflled at least one of the following exclusion criteria: (i) Non-English publications (ii) Systematic reviews Studies fulflling the criteria were included in a metaanalysis and meta-regression provided that they reported odds ratio (OR) on the risk of cataract in a sample of patients with asthma and/or COPD treated with inhaled and/or systemic corticosteroid.
A systematic search for recent literature was conducted (last updated March 2023) in the databases of PubMed and EMBASE to identify potentially eligible studies.
Te search string on PubMed consisted of the following text words: cataract and corticosteroids.
Te EMBASE search was conducted in the Ovid search database using a combination of the following text words "corticosteroid," "asthma," and "COPD" and the following subheadings "Diagnosis, Drug Terapy" of the subject heading "cataract."Te fnal search was conducted by combining all search concepts as the following: "corticosteroids or asthma or COPD and cataract/di, dt (Diagnosis, Drug Terapy)."

Study Selection and Data Extraction.
All studies retrieved from the searches in PubMed and EMBASE were screened by both authors (OS and CSU).Te fnal inclusion of a study in the present review was based on consensus between both authors.Studies were screened for eligibility based on the title and abstract.
Data from eligible studies on association between cataract development and corticosteroid treatment for asthma and/or COPD were obtained by the PICO approach for automatic data extraction.Both authors retrieved information from all studies fulflling the eligibility criteria and performed data extraction.Study details extracted included the following: study title and design, length of study, population, sample size, mean age of the included individuals, defnition of the control group, statistical analysis, outcomes reported, and results.
Te primary outcome of interest was the assessment of the cataract risk in the included studies.We also screened the studies for the applied defnition of asthma and/or COPD and whether included patients had been prescribed inhaled and/or systemic corticosteroid.We retrieved information on daily dose and whether oral corticosteroids were prescribed as a rescue course or as maintenance therapy.
For the development of a pooled efect size estimate in meta-analyses, only odds ratios were used as most of the eligible studies investigating the risk of cataract in asthma and/or COPD patients provided this efect size.Some studies, however, provided risk ratios or hazard ratios.Due to the inclusion of three types of observational studies, that is cohort, cross-sectional, and case-control studies, separate meta-analyses were performed according to the study type.

Quality Assessment.
Te frst author OS performed quality assessment of all included studies.Discrepancies were discussed afterwards between the authors and solved by consensus.All included studies were assessed individually for quality and risk of bias using the standard Newcastle-Ottawa Scale (NOS) for case-control and cohort studies [28] and the adapted NOS for cross-sectional studies [29].Te NOS consists of eight items within three categories.Te maximum total is nine, and a study score ≥7 is considered a high-quality study.Te adapted version used for the crosssectional studies consists of seven items within three categories with the maximum total score also being nine.

Data Analysis.
Statistical analyses were performed using the Meta-package for RStudio Version 1.2.5001 2009-2019 RStudio, Inc. Tis package includes the function MetaGen, which provides the generic inverse variance method for meta-analysis [30].Tis method is used for pooling of efect sizes such as the risk ratio or odds ratio in precalculated efect sizes based on binary outcome data (i.e., cataract or no cataract development).Te output is treatment estimates, standard errors, and confdence intervals as well as an estimate of statistical heterogeneity and tau 2 .Te package also includes the function MetaReg, which provides a method for meta-regression.Tis method incorporates a meta-analysis object and the input of covariate name such as a dichotomous outcome, that is dose-response or no doseresponse relationship between corticosteroid exposure and cataract risk in included studies.Sensitivity analysis was 2 Journal of Ophthalmology performed by excluding each study in a series to establish individual study impact on the pooled odds ratio (OR) estimates.An assessment of funnel plot asymmetry was tested using the Harbord test with a p value <0.10 signifying publication bias [31].

Characteristics of Included Studies.
Te search algorithm yielded a total of 2,793 hits.Te fowchart of the selection process of potentially eligible and included studies addressing a possible association between corticosteroid exposure in patients with asthma and/or COPD and cataract development is given in Figure 1.
A total of 19 studies fulflled all criteria and were included in the present systematic review, and furthermore, 12 studies (63%) were included in three separate meta-analyses of pooled estimates of odds ratios according to the study type.Studies included in a meta-analysis were either crosssectional or cohort studies except for two case-control studies.Two observational studies did not comprise efect sizes allowing pooling and were therefore not included in one of the meta-analyses [22,32].Te most common efect size in the included studies was OR, whereas fve studies reported risk ratios (RR) or hazard ratios (HR) and were, therefore, not included in the meta-analysis.
Table 1 summarizes the main characteristics of the studies.
In total, 1,274,878 individuals were included in this systematic review.Of these, 13,343 (1%) and 562,745 (44%), respectively, were included in the meta-analysis of crosssectional studies and cohort studies, and 25,693 (2%) were included in the meta-analysis of case-control studies.In the systematic review, there were 62,292 (5%) cases of cataract and 996,328 (78%) patients with asthma and/or COPD and prescribed corticosteroids.Of the total number of individuals included in the present systematic review, 642,621 (50%) COPD and/or asthma patients had ever or occasional (at least one 12 months before enrollment [17]) oral corticosteroid (OCS) prescription, 4,054 (0.3%) COPD and/or asthma patients prescribed a combination of both ICS and OCS during the treatment period, 290,891 (23%) COPD and/or asthma patients prescribed or had ever been prescribed ICS.
Four of the included observational studies comprised a cohort of cataract patients (n � 123,065) [25,26,34,37] and, similar to the other included studies, assessed the potential association with previous corticosteroid exposure.Te remaining studies included asthma or COPD patients prescribed corticosteroid either in a subgroup analysis or in the main population of the study.

Meta-Analysis.
In a pooled efect size estimate analysis of cohort studies, the weighted odds ratio applying a random efects model was 1.60 (95% CI 1.11 to 2.31).Similarly, the weighted odds ratio for cross-sectional and case-control studies were 2.78 (95% CI 2.09-3.71)and 2.0 (95% CI 0.93-4.30),respectively.More details of the fndings are given in Figure 2.   ).Only two case-control studies were included, and individual study's impact on the pooled OR estimates was therefore not assessed for this study design (Figure 3).

Quality Assessment.
Quality assessment and risk of bias of the included studies are illustrated in Tables 2 and 3.In the cross-sectional studies, there was a high risk of selection bias in half of the included studies as they provided no information on characteristics of the control group or the comparison group (Table 2).On the contrary, cohort and case-control studies had a low risk of bias for most categories presented in Table 3.

Publication Bias Assessment.
Included studies demonstrated small study efects as seen by the asymmetric funnel plot, which may be ascribed to publication bias.Te Harbord test was statistically signifcant (p < 0.01).Te trim and fll analysis found no evidence of publication bias (Figure 4).

Meta-Regression.
In a meta-regression of included studies dichotomized according to whether they reported a dose-response association between corticosteroid exposure and cataract development, a dose-response relationship was signifcantly associated with cataract development (OR 1.99, [95% CI 1.39 to 2.88], p < 0.001) (Figure 5).

Discussion
Te fndings in this systematic review of available studies suggest a strong association between ICS treatment and the risk of cataract in COPD and asthma patients, with the metaanalysis revealing on average a doubled risk for cataract development in corticosteroid-exposed COPD and asthma  patients and the meta-regression revealing a dose-response relationship between corticosteroid exposure and cataract risk.Eight studies reported an increased cataract risk in ICStreated asthma and COPD patients, and only two of these studies did not adjust for other corticosteroid exposure [21,22].Te increased cataract risk in COPD and asthma patients is, therefore, not likely infuenced by way of corticosteroid administration, and ICS treatment also increases cataract risk.
Almost all included studies using medical records and databases found a signifcant association between cataract risk and ICS exposure.In contrast, Miller et al. found no signifcant association between cataract risk and ICS treatment in COPD and were not able to assess the potential impact of corticosteroids administered during hospital admissions [36].Tey obtained information from GP's computerized medical records compiled into the General Practice Research Database (GPRD) [36].Derby et al., Jick et al., and Smeeth et al. on the other hand also used GPRD and reported a signifcant association between cataract and individuals with asthma and/or COPD prescribed corticosteroids [26,33,38].Tis was similar to the study by Derby and Miller who found that there was no increased cataract risk in patients treated with ICS [38].Te study by Jick et al. investigated ICS-treated patients according to daily prescribed doses of less than 500 μg to more than 1500 μg and found an increased cataract risk over a 12-month period [33], while Miller et al.only included patients receiving 10 or more prescriptions for ICS during the study period.Similar to the study by Jick et al., Smeeth et al. also found an increased risk (OR 1.55) of cataract [26].Tese results are consistent with our fndings, which indicated a doseresponse relationship between corticosteroid exposure and cataract risk.In addition, unlike the study by Miller et al. which only included COPD patients, all the other three studies included both asthma and COPD patients.Moreover, all included studies adjusted for age in their analysis and by that reduced the risk of confounding.Some studies also adjusted for treatment with systemic corticosteroid in order to reduce the risk of confounding and further focused on the impact of ICS on cataract development [25,26,34,35].
In the western world, most cases of COPD are caused by tobacco smoking [41].Besides corticosteroids, tobacco    Note."+" indicates a point."Ascertainment of exposure," "comparability," and "Assessment of outcome" are the only items with a scale of two points.Te remaining items are evaluated as zero or one point.

Journal of Ophthalmology
10 Journal of Ophthalmology  Note."+" indicates a point.A study can be given a maximum of one point for each item within the "Selection" and "Exposure" categories.A maximum of two points can be given for the "Comparability" category.
Journal of Ophthalmology smoke has also been suggested as a potential risk factor for cataract development [42,43].Studies included in the present review have also adjusted for smoking history in their analyses to reduce the risk of confounding [25,33,36,38].In addition, the underlying infammation in asthma has also been suggested as a risk factor for cataract development [44].A pathway possibly leading to the development of cataract in patients with allergic asthma has previously been described [44].However, the potential association between chronic airway infammation in asthma and the risk of cataract remains to be further investigated.
A systematic review and meta-analysis by Weatherall et al. looking at the association between corticosteroid exposure and cataract was published more than a decade ago [15].Te previous review focused on case-control studies investigating the association between ICS use and the cataract risk and found that use of ICS increased cataract risk by approximately 25% for each 1,000 μg increase in daily dose.Te current study also found clear support for an increased cataract risk with daily ICS dose exceeding 1,000 μg.In addition, our meta-regression confrmed a dose-response relationship between corticosteroid exposure and cataract risk.Te threshold of a daily ICS dose of at least 1,000 μg suggests a risk of cataract comparable to the risk caused by treatment with systemic corticosteroid [37,38].However, the study by Wang et al. showed that the risk of cataract development is almost twice as high in patients treated with systemic corticosteroids compared to patients only treated with ICS [21].Te previous systematic review and metaanalysis was, similar to our study, limited by statistical heterogeneity among the included studies in the analysis [15].Te study by Weatherall et al. did not provide assessment of study quality and bias risk due to the inclusion of only four studies.Given the objective of our study, and,  furthermore, the inclusion of a number of study designs, the present study adds important knowledge to the previously published study.In addition, a number of original studies reporting on the association between corticosteroid exposure and cataract development have been published in the last decade, which highlights the need for an update on the current understanding between cataract risk and corticosteroid exposure.
In general, most studies included could be rated as being of high quality (Table 2).However, a few limitations of the included studies are worth mentioning.Te study by Toogood et al. and Delcourt et al. did not report how asthma was defned [22,37].Te study by Derby and Maier, likewise, did not include a defnition of COPD [38].Some of the included studies based on information from medical records relied on a correct diagnosis of COPD and/or asthma and, perhaps, did not confrm diagnosis before enrollment.Te study by Cumming et al. and Wang et al. included asthma patients and obtained information from questionnaires regarding corticosteroid treatment for asthma or other chest diseases, which may have led to selection bias [21,25].Relatively few asthma subjects were included, which would have made it likely to detect an association between cataract and corticosteroid treatment [25].Other included studies used guideline recommendations for both corticosteroid treatment and diagnosis of asthma and COPD.For instance, the study by Ernst et al. used estimates of ICS exposure according to national asthma treatment strategies [34].Daugherty et al. and Sweeney et al. included patients with severe asthma defned according to the GINA [12,39].In addition, it was difcult to assess the overall exposure to both systemic and inhaled corticosteroids as studies only provided relatively few details on the treatment period.Studies either mentioned that patients were prescribed ICS regularly [22], daily [33,35,36] or currently [21,34], or had previously been prescribed OCS [37,38] or occasionally OCS [17] and did not specify whether it was maintenance treatment and/ or rescue courses.Te controls in the included studies in most studies had never been prescribed systemic corticosteroid, whereas patients were included provided they had ever had at least one prescription for either ICS or OCS.

Strengths and Limitations.
We identifed a large number of relevant studies in the past three decades and used, according to guidelines, two reviewers for the selection process, data extraction, and quality assessment of potentially eligible and fnally included studies.We provided, unlike the recent review on the association between corticosteroid exposure and cataract development, a metaregression with a dichotomous outcome, that is doseresponse or no dose-response association, and found signifcant indication of a dose-dependent association between corticosteroid exposure and the cataract risk [15].Te main limitation is that relatively few studies have been published within the last decade, while most of the studies investigating the possible association between corticosteroid exposure and the cataract risk are published some years ago.However, by including a number of study designs, we were able to include more studies and provide independent estimates of the cataract risk based on diferent populations compared to the previous review [15].In addition, studies were observational, and assessment of causality is, therefore, difcult, although the inclusion of these studies allowed a more thorough review of the existing literature.Te choice of study design could have led to limitations in our systematic review.Tough observational studies cover a wide range of studies where the disease of interest occurs spontaneously, they are more prone to bias and confounding.In a systematic review, randomized controlled trials (RCTs) are often preferred instead of observational studies, as they represent the highest level of evidence.Tey are, however, probably not the best for analyzing long-term side efects of drugs, also due to the majority of studies having relatively short duration.A systematic review and meta-analysis incorporating welldesigned observational studies with large populations and sufcient durations could be necessary for a more precise assessment of the relationship between corticosteroid exposure and cataract development.It is also necessary to reduce heterogeneity in methods used for cataract detection and corticosteroid exposure, including daily dose and duration of treatment.Our meta-analysis and meta-regression were limited by heterogeneity of included studies, especially concerning the methods used to detect cataract and the diferences in adjustment of variables in the statistical analyses.However, the use of a random efects model allowed for the efect size variation between studies to be taken into account.Furthermore, the asymmetry in study distribution in the funnel plot cannot rule out publication bias due to negative results being difcult to publish.In order to reduce the risk of false-positive results and an appearance of asymmetry in the funnel plot, we used the Harbord test instead of the Egger test as a statistical testing method for funnel plot asymmetry [45].Moreover, both publication bias assessment and the meta-regression analysis usually demand that the number of included studies exceeds ten.However, the number of included studies was too small for publication bias assessment and meta-regression analysis according to the study design.Both the publication bias assessment and the meta-regression analysis output should therefore be interpreted with caution.We performed both the publication bias assessment and meta-regression analysis by including all studies that reported similar efect measurements on target population and outcome, which could argue for the validity of both analyses.Also, similarly important, the trim and fll method analysis does not include other reasons than publication bias for the asymmetry of the funnel plot, and the method should be interpreted carefully when marked heterogeneity between the included studies exists [46].Lastly, the diferent profles of quality assessment evaluators may have caused difculties in detecting caveats in study methodology and could be the reason for the high scores on the quality assessment tools.Tis questions the accuracy of quality assessment and the possibility to clearly identify the risk of bias.However, the tools used for quality assessment and detection of bias risk are validated and provide a comprehensive and systematic assessment of the included studies.
Journal of Ophthalmology

Conclusion
We have identifed several studies revealing an association between corticosteroid exposure and cataract development in asthma and COPD patients.While the included studies mostly agree on the increased risk of cataracts in patients with COPD and asthma, there were biases and limitations associated with the included studies.Te most signifcant fndings are that many prescriptions and high daily dose of corticosteroid increase the risk of cataract signifcantly in both patients with asthma and COPD; that is, a daily dose of 1,000 μg or more of inhaled corticosteroids signifcantly increases the risk of cataract compared with the risk in patients exposed to a lower daily dose and that patients prescribed oral corticosteroids have a greater risk of cataract development compared to patients prescribed inhaled corticosteroids only.Te risk of cataract shown in our metaanalysis should always be weighed against the benefts of ICS in asthma and COPD patients.Cataract screening could, however, be applied for asthma and COPD patients prescribed high-dose ICS treatment in order to reduce the cataract development risk.More research is needed on the best-possible corticosteroid treatment strategies according to treatable traits and other disease characteristics in both asthma and COPD and also to reduce the future burden of cataract.
in the meta-analysis: 12

Figure 1 :
Figure 1: Consort diagram showing the study selection process.

Figure 2 :
Figure2: Forest plots depicting the fndings from the meta-analysis of (a) cohort, (b) cross-sectional, and (c) case-control studies.Odds ratio (OR) is illustrated as a vertical line.Te 95% confdence interval (CI) is illustrated as a line on both sides of the OR.Te heterogeneity test was completed to test for any diference between studies.A p value <0.05 was defned as statistically signifcant.

Figure 3 :
Figure3: Forest plots depicting the sensitivity analysis in (a) cohort, (b) cross-sectional, and (c) case-control studies.Summary estimates were calculated using a random efects model and by omitting one study at a time.Odds ratio (OR) is illustrated as a vertical line.Te 95% confdence interval (CI) is illustrated as a line on both sides of the OR.Te heterogeneity test was completed to test for any diference between studies.A p value <0.05 was defned as statistically signifcant.

Table 1 :
Studies examining the association between exposure of (a) inhaled corticosteroids (ICS), (b) oral corticosteroids (OCS), or (c) both ICS and OCS for asthma and/or COPD and the

Table 2 :
Quality and bias risk assessment in included cross-sectional studies.

Table 3 :
Quality and bias risk assessment in included cohort and case-control studies.
Figure 4: Funnel plot for the assessment of potential publication bias in studies addressing the association between corticosteroid exposure and cataract development.(a)Afunnelplot with no additional studies added.Harbod regression test: t � 3.55, df � 9, p value � 0.006.Including all studies revealed the following efect measurement using a random efects model: OR 2.01, with 95% CI � (1.53-2.63);p<0.001.(b)Represents the trim and fll analysis.Harbord regression test: t � 4.33, df � 9, p � 0.002.Trim and fll analysis: four studies added on the left side.Harbod regression test: t � −0.07, df � 13, p-value � 0.948.Te following efect measurement using a random efects model was revealed: OR 1.60, with 95% CI � (1.12-2.29);p�0.01.Note.Filled circles: observed fndings.Open circles: imputed and added studies after trim and fll analysis.Te study by Bloechliger et al. was omitted due to missing values.Figure5: Forest plot illustrating a meta-regression using a random efects model for the assessment of a dose-response association between corticosteroid exposure and cataract development.Included studies were dichotomized on whether they reported a dose-response association between corticosteroid exposure and cataract development.Sigma 2 � 0.003.Odds ratio (OR) is illustrated as a vertical box.Te 95% confdence interval (CI) is illustrated as a line on both sides of the OR.