Anterior chamber width (ACW) is defined as the horizontal scleral spur-to-scleral spur distance,10 while the lens vault (LV) is defined as the perpendicular distance between the anterior pole of the crystalline lens and the horizontal line joining the 2 scleral spurs.11 Anterior vault (AV) represents the sum of the LV and the anterior chamber depth (ACD). The anterior chamber area (ACA) is defined as the cross-sectional area of the anterior segment, bounded by the corneal endothelium, anterior surface of the iris, and anterior surface of the lens (within the pupil). A vertical axis through the midpoint of the ACA was plotted by the software, and the ACV was calculated by rotating the ACA 360° around this vertical axis.12 Angle opening distance at 750 µm (AOD750) is defined as the perpendicular distance from the iris to the trabecular meshwork at 750 µm anterior to the scleral spur.9 Trabecular iris space area at 750 µm (TISA750) refers to the trapezoidal area bordered by the anterior iris surface inferiorly, the inner corneoscleral wall superiorly, the AOD750 anteriorly, and posteriorly by a line perpendicular to the plane of the inner corneoscleral wall drawn from the scleral spur to the opposing iris.9 Iris area (IAREA) is the cumulative area of the entire iris, from the scleral spur to the pupil.6 Iris thickness 750 (IT750) and IT2000 are defined as the iris thicknesses at 750 µm and 2000 µm from the scleral spur, respectively. To calculate iris curvature (ICURV), the software draws a line from the most peripheral to the most central points of iris pigment epithelium and then a perpendicular line is extended from this line to the iris pigment epithelium at the point of greatest convexity.6 Posterior corneal arc length (PCAL) is defined as the arc distance of the posterior corneal border between the scleral spurs.13
AOD750 indicates angle opening distance at 750 µm. LV indicates lens vault.
eTable. Comparison of Baseline Variables In Participants Who Developed Gonioscopic Angle closure and Those With Open Angles and AS-OCT at Baseline and at 4 Years
Customize your JAMA Network experience by selecting one or more topics from the list below.
Nongpiur ME, Aboobakar IF, Baskaran M, et al. Association of Baseline Anterior Segment Parameters With the Development of Incident Gonioscopic Angle Closure. JAMA Ophthalmol. 2017;135(3):252–258. doi:10.1001/jamaophthalmol.2016.5847
Copyright 2017 American Medical Association. All Rights Reserved.
What quantitative anterior segment optical coherence tomography imaging parameters are associated with the development of incident gonioscopic angle closure at 4-year follow-up in eyes with gonioscopically open angles at baseline?
In this community-based observational study, a smaller angle opening distance at 750 µm and a larger lens vault were associated with development of incident gonioscopic angle closure at four year follow-up. These findings were statistically significant.
Anterior segment optical coherence tomography imaging parameters may be associated with the development of future incident gonioscopic angle closure. Angle opening distance 750 µm and lens vault measurements may serve as effective screening tools to identify eyes at risk of angle-closure glaucoma prior to the onset of a clinically significant disease.
Baseline anterior segment imaging parameters associated with incident gonioscopic angle closure, to our knowledge, are unknown.
To identify baseline quantitative anterior segment optical coherence tomography parameters associated with the development of incident gonioscopic angle closure after 4 years among participants with gonioscopically open angles at baseline.
Design, Setting, and Participants
Three hundred forty-two participants aged 50 years or older were recruited to participate in this prospective, community-based observational study. Participants underwent gonioscopy and anterior segment optical coherence tomography imaging at baseline and after 4 years. Custom image analysis software was used to quantify anterior chamber parameters from anterior segment optical coherence tomography images.
Main Outcomes and Measures
Baseline anterior segment optical coherence tomography measurements among participants with gonioscopically open vs closed angles at follow-up.
Of the 342 participants, 187 (55%) were women and 297 (87%) were Chinese. The response rate was 62.4%. Forty-nine participants (14.3%) developed gonioscopic angle closure after 4 years. The mean age (SD) at baseline of the 49 participants was 62.9 (8.0) years, 15 (30.6%) were men, and 43 (87.8%) were Chinese. These participants had a smaller baseline angle opening distance at 750 µm (AOD750) (0.15 mm; 95% CI, 0.12-0.18), trabecular iris surface area at 750 µm (0.07 mm2; 95% CI, 0.05-0.08), anterior chamber area (30 mm2; 95% CI, 2.27-3.74), and anterior chamber volume (24.32 mm2; 95% CI, 18.20-30.44) (all P < .001). Baseline iris curvature (−0.08; 95% CI, −0.12 to −0.04) and lens vault (LV) measurements (−0.29 mm; 95% CI, −0.37 to −0.21) were larger among these participants ( all P < .001). A model consisting of the LV and AOD750 measurements explained 38% of the variance in gonioscopic angle closure occurring at 4 years, with LV accounting for 28% of this variance. For every 0.1 mm increase in LV and 0.1 mm decrease in AOD750, the odds of developing gonioscopic angle closure was 1.29 (95% CI, 1.07-1.57) and 3.27 (95% CI, 1.87-5.69), respectively. In terms of per SD change in LV and AOD750, this translates to an odds ratio of 2.14 (95% CI, 2.48-12.34) and 5.53 (95% CI, 1.22-3.77), respectively. A baseline LV cut-off value of >0.56 mm had 64.6% sensitivity and 84.0% specificity for identifying participants who developed angle closure.
Conclusions and Relevance
These findings suggest that smaller AOD750 and larger LV measurements are associated with the development of incident gonioscopic angle closure after 4 years among participants with gonioscopically open angles at baseline.
Gonioscopy remains the reference standard for the diagnosis of angle closure and primary angle-closure glaucoma. However, this examination technique is highly subjective, as findings can be influenced by factors such as illumination or inadvertent pressure on the eye when using a gonioscopy lens. Agreement on gonioscopic findings is moderate even among experienced examiners.1-3
Anterior segment optical coherence tomography (AS-OCT) provides an objective alternative for examining anterior segment structures and the iridocorneal angle. We recently demonstrated that AS-OCT imaging predicts the future development of incident gonioscopic angle closure: individuals with quadrants of closed angles on baseline AS-OCT (but open gonioscopically) were more likely to develop incident gonioscopic angle closure after 4 years than individuals with open angles on baseline AS-OCT.4 However, it is unknown whether quantitative AS-OCT parameters, such as lens vault (LV), iris parameters, and angle opening distance (AOD), also predict the development of incident gonioscopic angle closure. Prior studies have demonstrated that these parameters are associated with angle closure, but these studies were cross-sectional in design.5-9
This study aims to identify quantitative AS-OCT imaging parameters that are associated with the development of incident gonioscopic angle closure after 4 years among participants with gonioscopically open angles at baseline.
This study adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review boards of the Singapore Eye Research Institute and Johns Hopkins University. All participants provided written informed consent and were compensated. The study population and methods have been previously described.4
A total of 2052 participants completed all baseline tests in 2007, of whom 1630 had open angles on gonioscopy. Of the 1630 individuals, 591 (36.3%) had angle closure in 1 or more quadrants on AS-OCT (defined as iris-trabecular contact beyond the scleral spur in a given quadrant). For the current study, we examined 485 of 591 (82.1%) randomly selected individuals (stratified by the number of quadrants of AS-OCT closure).4 We also examined 100 participants with open angles on both gonioscopy and AS-OCT; this sample was chosen by a computer-generated random selection from 738 eligible participants.4
After providing their medical and ophthalmic history, respondents underwent the following examinations at baseline and 4-year follow-up: visual acuity, AS-OCT imaging (Visante AS-OCT, version 184.108.40.206; Carl Zeiss Meditec), anterior chamber depth (ACD), and axial length measurements (IOLMaster, version 3.02; Carl Zeiss Meditec). All examinations were performed on 1 day. The AS-OCT and IOLMaster hardware and software versions used were the same at baseline and follow-up. Goldman applanation tonometry, gonioscopy (detailed later), and optic disc assessment using a 78-diopter (D) lens were also performed. Individuals were excluded at baseline and follow-up if they had a history of intraocular surgery, anterior segment laser treatment, or penetrating trauma to the eye. Additionally, individuals with aphakia or pseudophakia and those with corneal disorders that could influence AS-OCT imaging (eg, corneal endothelial dystrophy, opacification, and severe pterygium) were also excluded.
Anterior segment optical coherence tomography imaging was performed in dark room conditions (0 lux) by a single operator masked to the results of all other tests. The standard anterior segment single-scan protocol was used with scans centered at the pupil. To obtain the best-quality image, the examiner adjusted the saturation and noise and optimized the polarization for each scan. The examiner chose the best image with the least eyelid motion or image artifacts. The examiner was masked to all clinical data, including gonioscopic and baseline imaging results. Similar lighting conditions, scan protocols, and analysis methods were used for baseline and follow-up visits.
One cross-sectional horizontal AS-OCT scan of the nasal and temporal angle was evaluated for each participant (Figure 1). The Zhongshan Angle Assessment Program (Zhongshan Ophthalmic Center) was used to analyze the AS-OCT images for quantitative anterior segment measurements.10 The same version (release date July 20, 2007) was used for the baseline and follow-up image analyses. A single observer masked to clinical data analyzed deidentified images. For each image, the only observer input was identification of the 2 scleral spurs. The algorithm then automatically calculated the anterior segment parameters. The reproducibility of Zhongshan Angle Assessment Program–derived anterior segment measurements is very high, with an intraclass correlation coefficient more than 0.88.6,10,11
Anterior chamber width,12 LV13 anterior vault, anterior chamber area, anterior chamber volume (ACV)11 angle opening distance at 750 µm (AOD750), trabecular iris space area at 750 µm (TISA750),9 iris area, iris thicknesses at 750 µm and 2000 µm from the scleral spurs, iris curvature,6 and posterior corneal arc length14 for participants are illustrated in Figure 1 and defined in the legend.
Gonioscopy was performed in the dark in all participants by a single examiner (M.B.) masked to AS-OCT findings at baseline and follow-up. Nonindentation gonioscopy was performed using a Goldmann 2-mirror lens (Ocular instruments Inc) at high magnification (×16) with the eye aligned to the goniolens in the primary gaze position. Care was taken to avoid light falling on the pupil. A slight tilting of the gonioscopy lens was permitted in the presence of a convex iris profile. Indentation gonioscopy was also performed using a Sussman 4-mirror lens (Ocular instruments Inc). The angle in each quadrant was graded based on the anatomical structures observed during gonioscopy (grade 0 = no angle structures, grade 1 = Schwalbe line, grade 2 = anterior trabecular meshwork, grade 3 = posterior trabecular meshwork or scleral spur, and grade 4 = visible ciliary body). This grading system was derived from the modified Scheie angle grading system.6 A quadrant was considered “closed” if the posterior trabecular meshwork could not be seen in the primary position without indentation (grade 0, 1, or 2). Gonioscopic angle closure in an eye was defined as closure in 2 or more quadrants.
The IOLMaster measures ACD as the distance from the corneal epithelium to the anterior lens surface with lateral slit illumination. The mean of 5 readings obtained for ACD and axial length were used for subsequent analyses. All readings for each parameter were required to be within 0.05 mm of the reading within the highest signal to noise ratio. Goldmann applanation tonometry was used to measure intraocular pressure (IOP), and the vertical cup- disc ratio was determined clinically using a 78-D lens at the slitlamp (Haag-Streit Model BQ-900; Haag Streit). The median of 3 IOP readings was used for subsequent analyses. All clinical measurements were performed by a glaucoma fellowship-trained ophthalmologist (MB) masked to the baseline and follow up AS-OCT findings.
Statistical analyses were performed using SPSS statistics software (SPSS for Windows, version 20.0; IBM-SPSS). Only data from the right eye were analyzed in this study. To compare baseline variables among participants who developed incident gonioscopic angle closure at follow-up with those who did not, a 2-tailed t test was used for continuous variables and a χ2 test was used for categorical variables. An appropriate Bonferroni correction (α/12) was applied to correct for the multiple AS-OCT parameters, and the statistical significance for P values was set at .004. To identify the most important baseline risk factors associated with incident gonioscopic angle closure development at follow-up, we used a forward stepwise logistic regression model, including the most significant variable at each iteration. The odds ratio with 95% CI was calculated for the occurrence of gonioscopic angle closure per unit change in each significant parameter. The forward stepwise logistic regression model was repeated, with angle parameters excluded as one of the variables in consideration for the iteration. The area under the receiver operating characteristic curves (AUROCs) were calculated, along with sensitivity and specificity. Gonioscopic angle closure at the 4-year visit was used as the reference standard. Youden J index was used to identify optimal AS-OCT parameter cutoff points for predicting incident gonioscopic angle closure development. The AUROC was also estimated for the combined effect of AS-OCT parameters using the “angle closure score,” 15 calculated as follows: score = −28.986879 − 0.339910×(ACV) + 3.223506×(anterior chamber width) + 7.296654×(iris thickness 750) − 2.202824×(iris area) + 1.534522×(anterior chamber angle) + 0.003242×(LV).15 This score was validated in a Singaporean hospital-based study of patients with angle closure and population-based controls. It was derived from a logistic regression model that provided a shifted linear combination of 6 AS-OCT–based parameters.15 The AUROC for the combination of the most significant variables in the stepwise logistic regression was also calculated based on a shifted linear combination of the significant variables.
Telephone calls were made to 585 individuals (485 cases, 100 controls) to ask if they were willing to participate in follow-up examination. Of these individuals, 12 had died in the interim period, 30 developed new medical conditions such as heart attack or stroke that precluded examination, 60 could not be reached, and 118 declined follow-up. Of the 365 individuals recruited (62.4%), 297 were cases and 68 were controls. Twenty-three of these individuals had undergone bilateral cataract surgery and were no longer eligible, leaving 342 total participants. There were no proportional differences between cases and controls for study nonparticipation (188 [38.8%] vs 32 [(32.0%]; 6.80 %; 95% CI, −4.18 to 16.84; P = .21). Additionally, there were no differences between respondents and nonrespondents regarding race/ethnicity, axial length, baseline IOP, and gonioscopy grade, but there were fewer men among the respondents than nonrespondents (data not shown).4 Of the participants, 187 (54.7%) were women and 297 (86.8%) were Chinese. All participants had open angles on gonioscopy in 4 quadrants at baseline. At 4-year follow-up, 254 individuals (74.3%) had 0 quadrants of gonioscopic angle closure, 39 (11.4%) had 1 quadrant closed, 20 (5.8%) had 2 quadrants closed, 22 (6.4%) had 3 quadrants closed, and 7 (2.1%) had 4 quadrants closed. Thus, 14.3% of individuals developed 180° or more of angle closure, with an overselection of individuals with OCT evidence of angle closure at baseline being 4.4% among men and 9.9% among women. There were no participants who presented with or gave a history of episodes of acute angle closure. Primary angle-closure glaucoma was diagnosed in 4 participants, and 7 participants had a diagnosis of primary angle closure (3 categorized because of peripheral anterior synechia and 4 due to elevated IOP [≥21 mm Hg]).
There were no differences in age and race/ethnicity between individuals who developed incident gonioscopic angle closure at follow-up (defined as ≥2 quadrants closed, n= 49) compared with those with gonioscopically open angles on all quadrants at follow-up (defined as 0 quadrants closed, n = 254), though participants who developed incident gonioscopic angle closure were more likely to be women (18.2%; 95% CI, 2.11-32.08; P = .02) (Table 1). Participants who developed angle closure were also more likely to have the following baseline ocular characteristics: lower modified Scheie grade, shorter axial length, shallower ACD, smaller anterior chamber angle and ACV measurements, larger LV measurements, smaller AOD750 and TISA750, and larger iris curvature (all P < .004).
Broadening the definition of gonioscopically open angles to either 0 or 1 quadrants closed (n = 293) did not change the analysis results (data not shown). Similarly, restricting the definition of gonioscopic angle closure to 3 or more quadrants closed (n = 29) did not affect the analysis (data not shown).
We also performed a subanalysis to compare individuals with 0 quadrants closed on gonioscopy and AS-OCT at baseline and follow-up (n = 46) with individuals with 2 or more quadrant gonioscopic closures at follow-up (n = 49) (eTable in the Supplement). Similar to the previous analysis, baseline modified Scheie grade (0.41; 95% CI, 0.30-0.53), axial length (1.30 mm; 95% CI, 0.75-1.85), ACD (0.49 mm2; 95% CI, 0.36-0.61), anterior chamber area (5.44 mm2; 95% CI, 4.16-6.74), ACV measurements (46.29 mm2; 95% CI, 35.74-56.83), LV measurements (−0.46 mm; 95% CI, −0.61 to 0.32), AOD750 (0.31 mm; 95% CI, 0.26-0.36), TISA750 (0.15 mm; 95% CI, 0.13-0.18), and iris curvature (−0.14 mm; 95% CI, −0.19 to −0.09) were different between the 2 groups (all P < .004).
In the forward stepwise logistic regression model for baseline parameters with occurrence of gonioscopic angle closure as the dependent variable, the first iteration identified LV as the significant factor (odds ratio, 1.69; 95% CI, 1.43-1.99; P < .001) (Table 2). In the second iteration that included LV in the model, the next most significant factor was AOD750 (odds ratio, 3.27; 95% CI, 1.87-5.69; P < .001). This model consisting of LV and AOD750 explained 38% of the variance in angle closure occurrence, with LV accounting for 28% and AOD750 accounting for 10% of this variance. For every 0.1 mm increase in LV and 0.1 mm decrease in AOD750, the odds of developing gonioscopic angle closure were 1.29 (95% CI 1.07-1.57) and 3.27 (95% CI 1.87-5.69), respectively. In terms of per SD change in LV and AOD750, this translates to an odds ratio of 2.14 (95% CI, 2.48-12.34) and 5.53 (95% CI, 1.22-3.77), respectively. After excluding angle parameters (AOD and TISA) as variables in consideration for the iteration in the forward logistic regression model, LV still remained the most important variable associated with occurrence of angle closure. Anterior chamber volume was the next most significant factor, contributing to 6% of the variance.
We examined the sensitivity, specificity, and AUROC for baseline LV and AOD750 values, with 2 or more quadrants of angle closure defined as angle closure at the 4-year visit (Table 3). A baseline LV cut-off of 0.56 mm and an AOD750 cutoff of 0.3 mm or less were identified as the optimal cut-off values based on the Youden J index. In this post-hoc analysis aimed at maximizing the AUROC, an LV cutoff of 0.56 mm had 64.6% sensitivity, 84.0% specificity, and an AUROC of 0.76. The AOD750 cutoff of 0.3 mm or less had 95.8% sensitivity, 64.6% specificity, and an AUROC of 0.82 (Figure 2). The performance characteristics of the combined parameters for detecting angle closure at 4 years, as estimated by the “angle closure score”15 and the LV/AOD750 combination, was also calculated (Table 3). The AUROCs for this analysis were similar to that of AOD750 alone. There were also no differences in the AUROCs between parameters and between sex within each parameter.
We recently reported that the presence of AS-OCT angle closure at baseline is associated with the future development of incident gonioscopic angle closure.4 In this study, we extended this analysis to identify the quantitative AS-OCT imaging parameters associated with the development of incident gonioscopic angle closure. We have found that a larger LV and smaller AOD750 at baseline are associated with the development of gonioscopic angle closure 4 years later. Interestingly, women were also more likely to develop incident gonioscopic angle closure than men.
The LV is defined as the perpendicular distance between the anterior pole of the lens and the horizontal line joining the 2 scleral spurs. It is a measure of the amount of the lens located anterior to the anterior chamber angle. Previous studies have demonstrated that individuals with angle closure have a larger LV compared with controls, and that a larger LV increases risk for angle closure.13,16,17 Lens vault also helps predict variations in angle width and ACD.18,19 Importantly, the association of LV with angle closure is independent of other anterior segment parameters, including ACD and lens thickness.13,16 Our study provides additional perspectives on the importance of LV by demonstrating that it is also associated with the development of incident gonioscopic angle closure. On the exclusion of angle width parameters (AOD and TISA), LV remained the most significant variable associated with occurrence of angle closure.
Interestingly, a cross-sectional study of the entire study population at the time of baseline examination found that AOD750 was the most useful angle measurement for identifying individuals with gonioscopic angle closure.9 The optimal AOD750 cutoff was lower than in our study (approximately 0.24 mm vs 0.31 mm) but there was a similar AUROC (0.83). Given that we followed up the study participants prospectively over time and only a subset developed incident gonioscopic angle closure during this follow-up period, it is reasonable to expect the AOD750 cutoff to be higher than for individuals who already had gonioscopic angle closure at baseline. We also noted that the performance of the estimated “angle closure score” as well as the combination of the LV and AOD750 (the significant predictors of incident gonioscopic angle closure) was similar to that of AOD750 alone, with AUROCs of 0.82 to 0.83. Thus, AOD750 measurement on AS-OCT imaging may serve as a good rule-out test in community-based screening for angle closure.
This study has several limitations. First, baseline and follow-up examinations were performed in different locations, leading to variability in lighting conditions and changes in angle parameters because of differences in cross-sectional images obtained. Efforts were made to maintain similar low lighting conditions and center AS-OCT images at the pupil, though subtle differences may have remained. There were a small number of individuals, for instance, who had 1 or more quadrants closed on AS-OCT at baseline but no quadrant closure on AS-OCT at follow-up. Additionally, there were issues with identifying the scleral spur in a few images, and therefore data could not be analyzed. Improved imaging technology that enables 360° assessment of the anterior chamber angle rather than single cross-sectional images may help overcome this technical limitation. In addition, baseline and follow-up gonioscopy were performed by different individuals, which could have introduced some variability. However, most patients with open angles on gonioscopy and AS-OCT at baseline also had open angles at follow-up (49 patients, 75.4%), suggesting that bias toward diagnosing angle closure on follow-up gonioscopy is unlikely. Additionally, the response rate of 62.4% at follow-up may have introduced bias. While there were no statistically significant differences among respondents and nonrespondents for parameters such as race/ethnicity, axial length, baseline IOP, and modified Scheie grade, there were fewer men among the respondents than nonrespondents. Therefore, the overall incidence of angle closure detected is likely lower than the true incidence. Furthermore, individuals with worse cataracts may have been more likely to attend the screening, which may have biased the findings toward a higher incidence of angle closure. It is difficult to know how the lower response rate may have biased the association of other ocular parameters with incident angle closure. Additionally, it is likely that some eyes that remained gonioscopically open at 4 years would develop gonioscopic angle closure with longer follow-up. Lastly, the results of this study are only generalizable to individuals with incident gonioscopic angle closure, and the relevance of our findings to other clinical outcomes, such as acute angle-closure glaucoma, remain unknown.
To our knowledge, this is the first prospective clinical study to investigate quantitative AS-OCT imaging parameters associated with the development of incident gonioscopic angle closure among participants with open angles at baseline. Larger baseline LV and smaller AOD750 were strongly associated with the development of incident gonioscopic angle closure in this study. These measurements may serve as predictive parameters to identify eyes at risk of narrow angles and primary angle-closure glaucoma prior to the onset of clinically significant disease, enabling the identification of individuals that warrant closer follow-up and monitoring. Further studies are needed to determine whether these parameters also predict other clinical outcomes, including peripheral anterior synechia formation, acute angle closure attack, or chronic angle-closure glaucoma.
Corresponding Author: Tin Aung, FRCS (ED), PhD, Singapore National Eye Centre, 11 Third Hospital Ave, Singapore 168751 (email@example.com).
Accepted for Publication: December 14, 2016.
Published Online: February 9, 2017. doi:10.1001/jamaophthalmol.2016.5847
Author Contributions: Drs Baskaran and Nongpiur had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Nongpiur and Aboobakar are joint first authors.
Concept and design: Aboobakar, Narayanaswamy, Friedman, Aung.
Acquisition, analysis, or interpretation of data: Nongpiur, Aboobakar, Narayanaswamy, Sakata, Wu, Atalay, Friedman, Aung.
Drafting of the manuscript: Aboobakar, Atalay, Friedman.
Critical revision of the manuscript for important intellectual content: Nongpiur, Aboobakar, Narayanaswamy, Sakata, Wu, Atalay, Friedman, Aung.
Statistical analysis: Nongpiur, Aboobakar, Sakata, Atalay, Friedman.
Obtained funding: Narayanaswamy, Aung.
Administrative, technical, or material support: Wu.
Supervision: Friedman, Aung.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This study was supported by grant SHF/FG485S/2010 from the Singhealth Foundation, Singapore and the Singapore Translational Research Investigator Award (NMRC/STAR/0023/2014) from the Singapore Ministry of Health’s National Medical Research Council.
Role of the Funder/Sponsor: The Singhealth Foundation and the National Medical Research Council had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Create a personal account or sign in to: