Visual and anatomic outcomes of suprachoroidal hemorrhage: A systematic review and meta-analysis

Abstract

Topic:

To characterize the presentation, management, and outcomes of suprachoroidal hemorrhage (SCH) in published studies.

Clinical relevance:

SCH is a potentially devastating condition, but there is no high-quality evidence for the prognosis or management of SCH.

Methods:

We performed a systematic review and meta-analysis of peer-reviewed studies of SCH published in PubMed, EMBASE, Web of Science, or Google Scholar between January 1, 1990 and September 1, 2022. The protocol was prospectively registered on Open Science Framework (https://osf.io/69v3q/). Random-effects models were used to calculate the pooled estimate and 95% confidence intervals (95% CI) for visual acuity (VA) and anatomic outcomes. Uni variable and multivariable random-effects meta-regressions were performed to determine baseline factors associated with VA outcomes and anatomic success, defined as retina attached at last follow-up.

Results:

Sixty-eight studies comprising 1246 eyes of 1245 patients were included, with mean (SD) follow up of 14.0 (9.4) months. The pooled estimate (95% CI) for mean change in logMAR VA from baseline to last follow-up was −0.98 (−1.22, −0.74) (I²=88.4%), with 72.0% (63.5%, 80.5%) (I²=74.3%) achieving VA improvement ≥ 0.3 logMAR (3-line improvement in Early Treatment of Diabetic Retinopathy Study VA), 39.6% (32.5%, 46.7%) (I²=83.2%) achieving final VA 1.0 logMAR (Snellen equivalent 20/200) or better, and 75.5% (68.4%, 82.7%) (I²=74.7%) achieving anatomic success. In univariable meta-regression, studies with predominantly non-spontaneous SCH and with greater percent of eyes receiving systemic steroids were associated with greater improvement in logMAR VA, greater proportion of eyes with VA improvement ≥ 0.3 logMAR, and greater proportion of eyes achieving anatomic success (all P < 0.05). Studies with greater percent of eyes treated surgically were associated with greater proportion of eyes with VA improvement ≥ 0.3 logMAR in univariable and multivariable analysis (P < 0.05). The mean (SD) quality score across studies was 13.9 (2.3) out of 24, and all outcomes were of very low certainty of evidence.

Conclusion:

Although limited by heterogeneous observational studies, published reports of SCH indicate that most eyes with SCH experience some degree of VA improvement and anatomic success. However, final VA outcomes remain poor, with most cases resulting in severe visual impairment or blindness.

Suprachoroidal hemorrhage (SCH), an accumulation of blood in the suprachoroidal space, is an uncommon but potentially devastating complication that can occur during or after ocular surgery, trauma, or spontaneously. Massive SCH has been variably defined as extensive SCH, SCH causing expulsion of intraocular contents (“expulsive”), or SCH causing retinal apposition (“kissing”) and is associated with generally poorer outcomes.^1–3 While multiple theories exist on the precise pathophysiology of SCH, hypotony is commonly accepted as a major precipitating factor. Other risk factors for SCH include systemic risk factors such as advanced age, hypertension, atherosclerosis, diabetes, and coagulopathy, as well as ocular risk factors such as aphakia, glaucoma, myopia, prior ocular trauma, prior ocular inflammation, and prior SCH in the presenting or fellow eye.^1–3

Chu and Green reviewed cases of perioperative SCH published between 1970 and 1998 and found substantial variability in the incidence, risk factors, prophylaxis, management, and prognosis of SCH.¹ Since then, additional case reports, case series, and case-control studies on SCH have been published, but no systematic review and meta-analysis has been performed, and there remains no high-quality evidence to guide the prognosis or management of SCH. In particular, controversy remains over the need for and optimal timing of surgical intervention and specific surgical techniques. Some authors advocate for immediate drainage through sclerotomies performed at the time of intraoperative diagnosis of SCH,^4–7 while others advocate for delayed drainage either in the early postoperative period^7,8 or after waiting 7–14 days for clot lysis.^9–11

The purpose of this study is to perform a systematic review and meta-analysis of studies of SCH published between 1990 and 2022, with a particular focus on massive SCH, to determine the visual and anatomic outcomes and their association with presenting characteristics and management-related factors.

Methods

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Appendix A, available at https://www.ophthalmologyretina.org/).¹² This study adhered to the tenets of the Declaration of Helsinki and was exempt from the requirement for informed consent due to its retrospective nature. The search strategy and inclusion and exclusion criteria were specified prospectively, and the study protocol was prospectively registered on Open Science Framework (https://osf.io/69v3q/; Appendix B, available at https://www.ophthalmologyretina.org/).

Eligibility Criteria for Considering Studies for This Review

Studies were included if they met the following criteria: (1) peer-reviewed studies in the English language, (2) published between January 1, 1990 and September 1, 2022 to cover the most recent three decades, and (3) included at least three patients with suprachoroidal hemorrhage. Studies were excluded if they (1) were single case reports, (2) did not provide visual or anatomic outcomes, or (3) included the same cohort of patients as another publication, in which case the publication that reported clinical outcomes was included. We excluded single case reports to reduce the risk of bias from single cases and to allow for meta-regression of study-level data. We otherwise kept the inclusion criteria broad in order to maximize the number of studies we could analyze for this rare condition, and to allow for some variation in SCH characteristics and management which could be analyzed with meta-regression.

The primary outcome of the review was the mean change in logMAR (logarithm of the minimum angle of resolution) visual acuity (VA) from the diagnosis of SCH to the final follow-up. Secondary outcomes included the proportion of eyes achieving VA improvement of at least 0.3 logMAR (3-line improvement in Early Treatment of Diabetic Retinopathy Study (ETDRS) VA), the proportion of eyes achieving final VA of 1.0 logMAR (Snellen equivalent 20/200) or better, and the proportion of eyes achieving anatomic success, defined as retina attached at last follow-up. Studies had to report measurement of at least one of the outcomes to be eligible for inclusion, as the purpose of our meta-analysis was to examine outcomes of SCH.

Search Methods for Identifying Studies

A literature search was performed in PubMed, EMBASE, Web of Science, and Google Scholar for publications including the keywords “suprachoroidal hemorrhage”, “suprachoroidal haemorrhage”, “choroidal hemorrhage”, “choroidal haemorrhage”, “hemorrhagic choroidal detachment”, “haemorrhagic choroidal detachment”, “expulsive choroidal hemorrhage”, or “expulsive choroidal haemorrhage” published between January 1, 1990 and September 1, 2022 (Appendix C, available at https://www.ophthalmologyretina.org/). The last literature search was performed on September 10, 2022. Results of the literature search were managed with Microsoft Excel.

Study Selection

Two study authors (TL, AGE) independently performed the literature search and reviewed titles and abstracts to identify potential studies for inclusion. A final list of studies for full text review was determined through discussion with the senior author (BJK) and according to the pre-specified criteria listed above.

Data Collection and Risk of Bias Assessment

Data values to be extracted were specified prospectively and managed with Microsoft Excel. Two study authors (TL, AGE) independently reviewed the full text of all included studies and extracted data using two separate spreadsheets. Differences were adjudicated by a third study author (ZT) and the senior author (BJK). Data extracted included details about the study, patients, SCH characteristics, management, and visual and anatomic outcomes (Appendix D, available at https://www.ophthalmologyretina.org/). For case-control studies, data were only extracted for cases with SCH, not for controls. Data that were not explicitly provided were characterized as “not available”, and percentages were calculated with the denominator equal to the number of eyes for which information was available. We defined SCH as “massive” if it involved ≥ two quadrants, was macula involving, kissing, expulsive, or otherwise defined as “massive” by the original study authors. We considered cases of kissing SCH to involve the macula and four quadrants unless otherwise stated by the original text. Snellen VA data were converted to logMAR scale for statistical analysis. For low vision, we used a scale of count fingers (CF) = 2.0 logMAR, hand motion (HM) = 2.3 logMAR, light perception (LP) = 2.7 logMAR, and no light perception (NLP) = 3.0 logMAR based on prior studies utilizing the Freiburg Visual Acuity Test.^13,14 When the original study provided only logMAR VA without Snellen equivalent or specification of their scale for low vision logMAR values, we kept the logMAR values provided by the original study. We defined anatomic success as retina attached at last follow-up, as has been done in prior studies.^15–17

We assessed the quality and risk of bias of each study using a modified version of the RTI Item Bank on Risk of Bias and Precision of Observational Studies.¹⁸ The RTI Item Bank is designed as a comprehensive source of validated questions from which investigators can select specific items based on the needs of the review topic, and abbreviated versions have been used in systematic reviews before.¹⁹ We used a modified version of the instrument consisting of 8 questions, one for each of the 8 domains, to be more applicable to the format of studies included in our review (Appendix E, available at https://www.ophthalmologyretina.org/). Two study authors (TL, AGE) independently assessed the quality of each included study, and the final quality ratings were determined based on the mean of the two reviewers’ ratings.

Data Synthesis and Analysis

Open Meta-Analyst software (version for Windows 8, downloaded from http://www.cebm.brown.edu/open_meta/download.html) was used for the meta-analysis. Random-effects models were used to calculate the pooled estimate and its 95% confidence intervals (95% CI) for baseline characteristics and SCH visual and anatomic outcomes, which were presented as Forest plots. The percentage of total variation in the estimate of SCH outcomes across included studies due to heterogeneity was assessed using the I² statistic.²⁰ The I² ranges from 0% to 100% (0% represents no heterogeneity) with higher values representing larger heterogeneity. Sensitivity analyses were performed using leave-one-out meta-analysis to assess the robustness of pooled estimates of visual and anatomic outcomes. The metafor meta-analysis package for R was used to assess the risk of publication bias, which was performed using funnel plots, Egger’s test for funnel plot asymmetry, and the trim-and-fill method.^21–23 We performed random-effects meta-regression analyses to determine how the pre-specified presenting and management factors were associated with visual outcomes and anatomic success rates.²⁴ Mean follow-up duration (< 12 months versus ≥12 months) and study sample size (< 10 eyes versus ≥ 10 eyes) were included as factors in meta-regression to examine if either of these variables affected outcomes. Each factor was first evaluated using univariable meta-regression, and factors with P ≥ 0.05 were included into the multivariable meta-regression model. Significant associations were presented as scatter plots with best fit lines. The level of confidence in the body of evidence for each outcome was assessed using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) framework.²⁵

Results

Overview of Studies

A total of 3481 records were identified with literature search, of which 2081 unique records were screened, and 68 full-text studies were found to be eligible and included in this meta-analysis (Figure 1). The summary statistics and meta-analysis estimates are provided in Table 1, and baseline characteristics and outcomes for each included study are shown in Table S2 (available at https://www.ophthalmologyretina.org/). The 68 studies (51 case series, 17 case-control studies) comprised 1246 eyes of 1245 patients. The mean (SD) age of patients across studies was 62.7 (16.4) years. The estimate (95% CI) for the percent of male patients across studies was 48.0% (42.1%, 53.8%).

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines diagram.

Table 1.

Summary of study-level data.

Measure	Value or Estimatea	N Studiesb	N Eyesb
Demographics
N patients	1245	68	1246
N eyes	1246	68	1246
Mean age (years): mean (SD), [range]	62.7 (16.4), [4.6, 82.0)	64	1050
Gender, male, % (95% CI)	48.0 (42.1, 53.8)	60	1054
SCH Characteristics
Precipitating event type, % (95% CI)
Perioperative	89.6 (87.1, 92.1)	68	1246
Traumatic	5.5 (3.8, 7.3)	68	1246
Spontaneous	5.3 (3.6, 7.1)	68	1246
Mean number of quadrants, mean (SD), [range]	3.7 (0.5), [2.2, 4.0]	27	454
Macula involving, % (95% CI)	74.6 (65.8, 83.4)	31	407
Kissing, % (95% CI)	66.4 (55.4, 77.5)	36	640
Expulsive, % (95% CI)	33.4 (22.0, 44.8)	30	453
Massive, % (95% CI)	75.3 (68.6, 82.1)	54	950
Management and Outcomes
Surgical treatment, % (95% CI)	64.4 (54.7, 74.0)	64	1082
Mean time until surgery in days, mean (SD), [range]	10.9 (9.8), [0, 40.2]	42	650
Use of systemic steroids, % (95% CI)	48.9 (25.4, 72.3)	15	147
Mean follow up in months, mean (SD), [range]	14.0 (9.4), [0.8, 47.4]	49	596
Mean LogMAR VA pre-SCH, mean (SD), [range]	1.28 (0.58), [0.44, 2.43]	18	286
Mean LogMAR VA pre-treatment, mean (SD), [range]	2.45 (0.34), [1.30, 3.00]	36	463
Mean LogMAR VA post-treatment, mean (SD), [range]c	1.59 (0.61), [0.17, 3.05]	53	696
Mean change in LogMAR VA from pre-treatment to post-treatment, estimate (95% CI)d	−0.98 (−1.22, −0.74)	29	227
Eyes with VA improvement ≥ 0.3 logMAR from pre-treatment to post-treatment, % (95% CI)d	72.0 (63.5, 80.5)	32	237
Eyes with final logMAR VA 1.0 or better, % (95% CI)c	39.6 (32.5, 46.7)	57	874
Eyes with anatomic success, % (95% CI)	75.5 (68.4, 82.7)	40	342

CI = confidence interval; logMAR = logarithm of the minimum angle of resolution; N = number; SCH = suprachoroidal hemorrhage; SD = standard deviation; VA = visual acuity.

^aFor percentages, presented as estimate (95% confidence interval) based on study-level meta-analysis.

^bNumber of studies or eyes for which information was available.

^cSome studies reported mean final VA without providing patient-level data to calculate the percentage of patients achieving final logMAR VA 1.0 or better. Other studies reported the percentage of patients achieving final logMAR VA 1.0 or better without reporting the overall mean final VA. As a result, the number of studies for these two outcomes differs.

^d29 studies reported both mean and standard deviation for change in logMAR visual acuity, which are both required for meta-regression of continuous outcome variables. 32 studies reported change in logMAR visual acuity with sufficient patient-level detail to determine the percentage of eyes that achieved VA improvement ≥ 0.3 logMAR. As a result, the number of studies for these two outcomes differs.

Clinical Characteristics and Management

Across studies, the estimate (95% CI) for type of SCH was 89.6% (87.1%, 92.1%) perioperative, 5.5% (3.8%, 7.3%) traumatic, and 5.3% (3.6%, 7.1%) spontaneous. Of the 63 studies that included eyes with perioperative SCH, the index surgery associated with development of SCH included 441 (39.8%) cataract or lens surgeries, 349 (31.5%) glaucoma surgeries, 272 (24.5%) vitreoretinal surgeries, 43 (3.9%) corneal surgeries, and 3 (0.3%) other surgeries (1 radiation plaque, 2 not reported). The mean (SD) number of quadrants involved was 3.7 (0.5). The estimate (95% CI) for percent of eyes was 74.6% (65.8%, 83.4%) macula involving, 66.4% (55.4%, 77.5%) kissing, 33.4% (22.0%, 44.8%) expulsive, and 75.3% (68.6%, 82.1%) massive.

The estimate (95% CI) for percent of eyes managed surgically was 64.4% (54.7%, 74.0%), with mean (SD) time until surgery of 10.9 (9.8) days. The estimate (95% CI) for percent of eyes managed with systemic steroids was 48.9% (25.4%, 72.3%). Fifteen studies reported information on the use of systemic steroids; of the nine studies in which at least one patient received systemic steroids, three studies used oral steroids, one study used intravenous followed by oral steroids, and five studies did not specify whether oral or intravenous systemic steroids were used. The mean (SD) follow-up after diagnosis of SCH was 14.0 (9.4) months.

Visual and Anatomic Outcomes

Across all studies, the mean (SD) VA was 2.45 (0.34) logMAR (Snellen equivalent HM to LP) at the time of SCH diagnosis and 1.59 (0.61) logMAR (Snellen equivalent 20/778) at last follow-up. The pooled estimate (95% CI) for mean change in logMAR VA from time of SCH diagnosis to last follow-up was −0.98 (−1.22, −0.74) (Figure 2A; I²=88.4%, P < 0.01). The estimate (95% CI) was 72.0% (63.5%, 80.5%) for achieving improvement in VA of at least 0.3 logMAR from SCH diagnosis to last follow-up (3-line improvement in ETDRS VA) (Figure 2B; I²=74.3%, P < 0.001), 39.6% (32.5%, 46.7%) for achieving final VA 1.0 logMAR (Snellen equivalent 20/200) or better (Figure 2C; I²=83.2%, P < 0.001), and 75.5% (68.4%, 82.7%) for achieving anatomic success (Figure 2D; I²=74.7%, P < 0.001).

Figure 2A.

Forest plot of change in logMAR visual acuity.

The mean VA prior to development of SCH was only reported in 18 studies (286 eyes) and was 1.28 (0.58) logMAR (Snellen equivalent 20/381); studies that reported mean VA prior to development of SCH had a significantly different distribution of precipitating surgeries than studies that did not (P < 0.001, chi-squared test), with more glaucoma-associated cases (48.0% vs 26.1%) and less cataract/lens-associated cases (26.6% vs 44.1%). Among 17 studies (267 eyes) that reported both mean pre-SCH VA and mean final VA, the mean (SD) pre-SCH VA was 1.25 (0.58) logMAR (Snellen equivalent 20/356), the mean (SD) final VA was 1.74 (0.74) logMAR (Snellen equivalent 20/1099), and the mean (SD) change in VA from pre-SCH to last follow-up was 0.50 (0.68) logMAR (5-line decrease in ETDRS VA). The primary and secondary visual and anatomic outcomes did not differ significantly between 18 studies that reported pre-SCH VA and 50 studies that did not (P > 0.05 for all outcomes, univariable random-effects meta-regression).

Sensitivity analyses performed using the leave-one-out method did not reveal significantly different estimates for any of the visual or anatomic outcomes. In funnel plots for assessing the publication bias, no significant funnel plot asymmetry was found for mean change in logMAR VA (P = 0.62), achieving improvement in VA ≥ 0.3 logMAR (P = 0.28), or achieving final VA of 1.0 logMAR or better (P = 0.34). However, significant funnel plot asymmetry was found for log odds of eyes achieving anatomic success (P = 0.005), suggesting risk of reporting bias, with an updated effect estimate (95% CI) from the trim-and-fill method of 67.5% (58.8%, 75.2%) for achieving anatomic success (Figure S3, available at https://www.ophthalmologyretina.org/).

Results from Meta-Regression for Factors Associated with Visual and Anatomic Outcomes

In univariable meta-regression, studies with predominantly spontaneous SCH were associated with worse visual and anatomic outcomes than studies with predominantly perioperative or traumatic SCH. Specifically, studies with predominantly spontaneous SCH had less improvement in mean logMAR VA (−0.09) than studies with predominantly perioperative SCH (−1.03, P = 0.04, Table 3), and were associated with a lower proportion of eyes achieving VA improvement ≥ 0.3 logMAR (32% vs 75% (perioperative) and 89% (traumatic), P = 0.009, Table 4) or anatomic success (10% vs 79% (perioperative) and 89% (traumatic), P < 0.001, Table S5, available at https://www.ophthalmologyretina.org/).

Table 3.

Univariable meta-regression for factors associated with mean change in logMAR visual acuity.

Factor	N Studies	N Eyes	Mean (SE)	Difference (SE)	P-value
Study decade					0.69
1990–1999	9	65	−1.05 (0.19)	Reference
2000–2010	6	40	−0.76 (0.22)	0.28 (0.37)	0.44
2011–2022	14	122	−1.03 (0.15)	0.02 (0.30)	0.96
Predominant SCH type					0.10
Perioperative	25	209	−1.03 (0.10)	Reference
Traumatic	2	7	−1.23 (0.41)	−0.20 (0.50)	0.69
Spontaneous	2	11	−0.09 (0.36)	0.93 (0.45)	0.04
Percent massive SCHa	27	208		0.006 (0.006)	0.32
Percent managed surgicallya	29	227		−0.012(0.014)	0.41
Mean days until surgery (per day increase)	25	200		−0.007 (0.015)	0.62
Percent use of systemic steroidsa	9	76		−0.009 (0.004)	0.02
Mean follow-up duration					0.31
<12 months	12	87	−1.16 (0.16)	Reference
≥12 months	15	121	−0.79 (0.14)	0.37 (0.26)	0.15
Unknown	2	19	−1.22 (0.38)	−0.06 (0.51)	0.91
Study sample size
<10 eyes	19	102	−0.89 (0.13)	Reference
≥10 eyes	10	125	−1.13 (0.17)	−0.25 (0.27)	0.35

LogMAR = logarithm of the minimum angle of resolution; N = number; SCH = suprachoroidal hemorrhage; SE = standard error.

^aPer one percent increase.

Table 4.

Univariable and multivariable meta-regression of factors associated with proportion of eyes with visual acuity improvement ≥ 0.3 logMAR.

				Univariable Meta-Regression		Multivariable Meta-Regression
Factor	N Studies	N Eyes	Mean (SE)	Difference (SE)	P-value	Difference (SE)	P-value
Study decade					0.65
1990-1999	10	67	0.66 (0.08)	Reference
2000-2010	6	40	0.71 (0.10)	0.048 (0.131)	0.71
2011-2022	16	130	0.76 (0.06)	0.094 (0.101)	0.35
Predominant SCH type					0.009		0.36
Perioperative	27	216	0.75 (0.04)	Reference		Reference
Traumatic	2	7	0.89 (0.16)	0.139(0.166)	0.40	0.093 (0.153)	0.54
Spontaneous	3	14	0.32 (0.14)	−0.433 (0.149)	0.004	−0.214 (0.159)	0.18
Percent massive SCHa	29	213		−0.002 (0.002)	0.31
Percent managed surgicallya	32	237		0.008 (0.002)	<0.001	0.006 (0.002)	0.003
Mean days until surgery	25	200		0.001 (0.005)	0.88
Percent use of systemic steroidsa,b	9	76		0.003 (0.001)	0.01
Mean follow-up duration					0.18
<12 months	13	92	0.77 (0.07)	Reference
≥12 months	15	121	0.74 (0.06)	−0.029 (0.091)	0.75
Unknown	4	24	0.51 (0.12)	−0.254(0.139)	0.07
Study sample size
<10 eyes	22	113	0.68 (0.06	Reference
≥10 eyes	10	124	0.79 (0.07)	0.110(0.092)	0.23

LogMAR = logarithm of the minimum angle of resolution; N = number; SCH = suprachoroidal hemorrhage; SE = standard error.

^aPer one percent increase.

^bPercent use of systemic steroids was not included in the multivariable meta-regression due to the small number of studies.

The percent of eyes with massive SCH was associated with a lower proportion of eyes achieving final VA of 1.0 logMAR or better, with the strength of the association approaching statistical significance in univariable meta-regression (P = 0.06, Table S6, available at https://www.ophthalmologyretina.org/).

The percent of eyes managed surgically was significantly positively associated with the proportion of eyes with VA improvement ≥ 0.3 logMAR (P < 0.001, Table 4, Figure S4, available at https://www.ophthalmologyretina.org/) but not with other outcomes in univariable meta-regression. This association remained significant in multivariable meta-regression (P = 0.003, Table 4). For eyes managed surgically, the mean number of days between SCH and surgical intervention (with 0 days representing immediate surgical intervention) was not significantly associated with any outcomes in univariable or multivariable meta-regression (Tables 3–4, Tables S5–S6, available at https://www.ophthalmologyretina.org/).

In univariable meta-regression of 15 studies with information on use of systemic steroids, a higher percent of eyes managed with systemic steroids was significantly associated with greater mean improvement in logMAR VA (P = 0.02, Table 3, Figure S5, available at https://www.ophthalmologyretina.org/), greater proportion of eyes with VA improvement ≥ 0.3 logMAR (P = 0.01, Table 4, Figure S6, available at https://www.ophthalmologyretina.org/), and greater proportion of eyes with anatomic success (P = 0.04, Tables S5, Figure S7, available at https://www.ophthalmologyretina.org/). Because information on use of systemic steroids was only available for 15 of 68 studies, the sample size was insufficient to examine in multivariable meta-regression.

Study decade, mean follow-up duration (< 12 months versus ≥12 months), and study sample size (< 10 eyes versus ≥10 eyes) were not significantly associated with any outcomes in univariable or multivariable meta-regression (Tables 3–4; Tables S5–6, available at https://www.ophthalmologyretina.org/).

Risk of Bias and Grading of Recommendations, Assessment, Development, and Evaluation

Out of a maximum total of 24 possible points in quality score, the mean (SD) study quality score was 13.9 (2.3) and ranged from 9 to 19.5. Quality scores were limited by the retrospective nature of studies (62/68 studies) and lack of an appropriately balanced comparison group (37/68 studies without a comparison group, 15/68 studies with a comparison group that was not appropriately balanced with the treatment group). In contrast, most (37/68) studies had sufficient follow up (≥ 3 months) with similar follow up durations across comparison groups (Table S7, available at https://www.ophthalmologyretina.org/).

All outcomes of meta-analysis and meta-regression were considered to be of very low certainty of evidence, due to reliance on observational studies. The level of certainty was downgraded for inconsistency due to substantial levels of heterogeneity for all outcome measures, and for potential publication bias with anatomic success outcomes (Table S8, available at https://www.ophthalmologyretina.org/).

Discussion

SCH remains a potentially devastating complication with limited high-quality evidence to guide management. We performed a systematic review and meta-analysis of 68 published studies of SCH and found that a majority of eyes with SCH experience some degree of VA improvement and anatomic success, but final VA outcomes remain poor, with a majority of cases resulting in severe visual impairment or blindness. To our knowledge, these findings represent the largest systematic review and meta-analysis of published studies of SCH to date.

While our understanding of SCH and surgical techniques and technology for SCH has improved over the past three decades, our findings show that visual prognosis after SCH remains guarded. While a majority of eyes in our analysis obtained a three-line improvement in ETDRS VA or achieved anatomic success, less than 40% achieved a final VA of 20/200 or better, meaning that a majority of SCH cases had severe visual impairment or blindness based on World Health Organization definitions.²⁶ This is in line with outcomes reported in prior large-scale case series (>50 cases), which have reported rates of achieving final visual acuity of 20/200 or better ranging from 15.7% to 60%.^27–29 Outcomes are worse when examining eyes with massive SCH or concurrent retinal pathology. For example, Ling et al. found that all cases that underwent attempted intraoperative drainage sclerostomy for massive SCH or secondary posterior segment interventions for retinal detachment resulted in VA worse than 20/200 at 6 months.²⁹ Furthermore, publication decade was not a significant predictor of visual or anatomic outcomes in our meta-regression, suggesting that improved surgical technology and techniques have not translated into better outcomes for SCH over time.

It is important to recognize that many eyes that develop SCH may have pre-existing ocular pathology that may limit their visual potential improvement independent of the course or management of SCH. The mean VA prior to development of SCH in our study was 1.28 logMAR (Snellen equivalent 20/381), although this was only reported in 18 studies, among which SCH associated with glaucoma surgery was disproportionately represented. This highlights the importance for future studies of SCH to report pre-SCH VA, so that visual and anatomic outcomes can be interpreted in light of baseline VA.

Surgical management of SCH has been advocated as one way to improve outcomes, although there remains no randomized trial data or high-quality evidence to guide surgical management. Our analysis found that the percent of eyes managed surgically was significantly positively associated with the proportion of eyes achieving VA improvement ≥ 0.3 logMAR in univariable and multivariable analysis, but not with other measures of visual or anatomic outcomes. Few prior studies of SCH have directly examined the effect of surgical management on SCH outcomes. Some observational studies have found no significant difference between patients undergoing surgical drainage and those who did not, or between patients undergoing external drainage alone versus combined drainage and PPV, but these conclusions are limited by small sample sizes and potential underlying differences in comparison groups.^15,16 Our analysis also found that the timing of surgery was not significantly associated with any outcome measures. Some authors advocate for immediate drainage of SCH at the time of diagnosis,^4–7 while others advocate for waiting 1–2 weeks until lysis of the suprachoroidal clot can be confirmed echographically.^9–11 Commonly cited indications for early SCH drainage include retinal apposition, macular involvement, extension to more than two quadrants posterior to the equator, or presence of concurrent retinal tear or detachment, but no prospective studies have been conducted for assessing whether early surgical intervention for these indications result in better outcomes.^9,28,30 Analysis of patient-level data from the SCH literature may elucidate whether early surgical management is associated with better outcomes, and it may provide further understanding of potential benefits of surgical intervention.

Interestingly, among 15 studies with information on use of systemic steroids, we found that the percent of eyes managed with systemic steroids was significantly associated with greater mean improvement in logMAR VA, greater proportion of eyes with VA improvement ≥ 0.3 logMAR, and greater proportion of eyes with anatomic success in univariable analyses. The effect of systemic steroids on visual outcomes was not directly examined in any of the studies included in our meta-analysis, and there are no randomized trials or high-quality evidence to guide the use of steroids in the management of SCH. SCH can be expected to cause a significant degree of intraocular inflammation, and there may be a theoretical benefit to aggressive reduction of inflammation. Our analysis suggests that use of systemic steroids was associated with better outcomes, although this should be interpreted cautiously given the small number of studies with details on steroid use, which prevented us from evaluating its independent effect in multivariable meta-regression.

We also found that studies with predominantly spontaneous SCH had significantly worse visual and anatomic outcomes than studies with predominantly perioperative or traumatic SCH. Our meta-analysis included three studies with predominantly spontaneous SCH encompassing 14 eyes.^31–33 These included three eyes from patients on anti coagulation with warfarin, two eyes from one patient with hypertension and bilateral age-related macular degeneration, three eyes that developed expulsive SCH in the setting of spontaneous corneal perforation, and six eyes with increased axial length that developed spontaneous concurrent SCH and rhegmatogenous retinal detachment. Due to the relative rarity of spontaneous SCH compared to perioperative and traumatic SCH, there are few published studies of spontaneous SCH, and it is difficult to say with confidence whether spontaneous SCH is truly associated with poorer outcomes or if the eyes included in these three studies had poorer outcomes due to chance or reporting bias. Possible explanations for why spontaneous SCH may be associated with poorer outcomes include underlying risk factors that predispose to more severe SCH (e.g., anticoagulation, choroidal vascular fragility in eyes with axial myopia) or delayed presentation relative to perioperative or traumatic cases (two cases described by Zhang et al. presented seven and ten days after symptom onset, and the three cases presented by Oshida et al. occurred in elderly disabled patients living in nursing homes).^32,33

While our study represents the largest systematic review and meta-analysis of published studies of SCH to date, some limitations of our study should be noted. First, our study was a meta-analysis of case series and case-control studies of SCH. We found no prospective randomized trials of SCH published during the time period of interest, so the studies included in our analysis are subject to the biases observed with non-randomized studies. Specifically, studies included in this meta-analysis were mostly retrospective and did not include appropriately balanced comparison groups. Furthermore, published studies of SCH may preferentially report on patients with severe or notable clinical courses or who were treated with a particular technique. Therefore, there may be an element of ascertainment bias, and our findings may not be representative of the population of all patients who develop SCH, although funnel plot analysis only identified significant risk of publication bias for one of the four outcomes (anatomic success). Second, not all studies provided sufficient information for our variables of interest, so certain analyses were performed using a subset of studies, limiting the power of meta-regression. Third, substantial heterogeneity was seen across studies for many of the variables we examined, potentially due to individual studies having relatively few numbers of eyes, which may have affected the precision of our meta-analysis estimates, although sensitivity analysis performed using the leave-one-out method did not reveal significantly different estimates for any of the outcomes. Finally, this analysis was performed using study-level metrics rather than eye-level data, so conclusions can only be drawn about relationships across studies rather than for individual eyes. This also precluded us from examining the effects of eye-level factors such as type of precipitating surgery or specific surgical intervention, as it was difficult to perform meta-regression for these factors at the study level with sufficient sample size. However, since approximately one-quarter of studies included in this analysis did not provide eye-level data, our analysis allowed for an inclusion of a larger and more representative sample of published studies.

In conclusion, findings from this systematic review and meta-analysis of studies of SCH published in the last three decades suggest, with the limitations stated above, that visual prognosis after SCH remains guarded, with a majority of cases resulting in severe visual impairment or blindness. This highlights the importance of conducting prospective studies and continuing efforts to develop novel management approaches and technologies that could potentially improve outcomes for this devastating condition.

Figure 2B.

Forest plot of proportion of eyes with visual acuity improvement ≥ 0.3 logMAR.

Figure 2C.

Forest plot of proportion of eyes with final visual acuity of 1.0 logMAR (Snellen 20/200) or better.

Figure 2D.

Forest plot of proportion of eyes with anatomic success, defined as retina attached at last follow-up.

This meta-analysis of 68 studies, supported by very low certainty evidence, found that most eyes with suprachoroidal hemorrhage experience some visual improvement and anatomic success, but most cases result in severe visual impairment or blindness.

Abbreviations:

CF

count fingers

CI

confidence interval

ETDRS

Early Treatment of Diabetic Retinopathy Study

HM

hand motion

logMAR

logarithm of the minimum angle of resolution

LP

light perception

N

number

NAV

not available

NLP

no light perception

SCH

suprachoroidal hemorrhage

SD

standard deviation

SE

standard error

UK

United Kingdom

USA

United States of America

VA

visual acuity