Introduction
Ocular complications of diabetes, including diabetic retinopathy (DR) and diabetic macular edema (DME), remain significant causes of blindness despite substantial therapeutic advances in recent years. Diabetes may impact the nature of the vitreomacular interface (VMI), and stronger adhesions of the vitreous to the retina in diabetes are linked to late complications such as tractional retinal tears, retinal detachment, and loss of vision.1 But its role in the development and progression of earlier stages of DR and DME are still unclear.2 Complete posterior vitreous detachment (PVD) may mitigate proliferative DR,3 owing to lack of a suitable substrate for neovascular growth, reductions in tangential and antero-posterior tractional forces, and increased oxygen tension in liquefied vitreous.4,5 Acting through similar mechanisms, the presence of a PVD may be associated with a reduced risk of macular edema.6,7
The studies that led to these findings were limited by small sample sizes, but also by use of older methods to detect PVD. Visualization of the VMI can be achieved through a variety of methods including B-scan ultrasonography, slit-lamp biomicroscopy, and optical coherence tomography (OCT).8 As OCT technology has advanced from time-domain (TD-OCT) to spectral-domain (SD-OCT), image resolution has greatly improved. Although TD-OCT has been used to evaluate the VMI and its effect on DME,9 SD-OCT provides far greater resolution of vitreous details and has not yet been utilized to study the relation of PVD to DME, to the best of our knowledge. SD-OCT permits visualization of the bursa premacularis, vitreous opacities, and posterior vitreous cortex — features that are often missed by other diagnostic modalities.10,11 In this study, we aimed to determine whether a relationship exists between PVD, as detected by SD-OCT, and incidence of treatable DME.
Patients and Methods
This multicenter, retrospective, observational study included consecutive patients with diabetes who underwent macular SD-OCT imaging in the Department of Ophthalmology at Saint Louis University and the Department of Ophthalmology at Washington University in St. Louis from January 1, 2012, to June 1, 2017. It was approved by both the Saint Louis University and Washington University institutional review boards. This study was performed in accordance with the Declaration of Helsinki and was Health Insurance Portability and Accountability Act compliant.
Eligibility and Exclusion Criteria
Patients were eligible if they had a diagnosis of diabetes mellitus, had a full-volume macular SD-OCT scan (patients with only single B-scan images were not eligible), and did not have proliferative diabetic retinopathy (PDR), DME, or overt vitreomacular traction (VMT) at the initial visit. Patients were excluded if they were missing glycated hemoglobin (HbA1c) or diabetes data or had a history of previous ocular surgery (except cataract surgery) or prior treatment for DR or DME (including prior intravitreous injection or laser). Patients not receiving DME treatment were excluded if they had less than 2 years follow-up or had only one encounter. However, patients with less than 2 years' follow-up were not excluded if they received DME treatment after the initial visit.
Study Protocol
We categorized PVD using SD-OCT imaging performed at the initial encounter. To make such assessments, we reviewed macular cube and raster scans on Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) and Cirrus HD-OCT (Zeiss, Jena, Germany). We then ranked PVD status according to criteria previously described by Johnson (Figure 1).12
Classification of PVD
The various stage classifications of PVD are as follows. Stage 0: No evidence of PVD; Stage 1: Perifoveal vitreous detachment with vitreofoveal adhesion; Stage 2: Perifoveal vitreous detachment with no vitreofoveal adhesion; Stage 3: Near-complete PVD except for vitreopapillary adhesion; and Stage 4: Complete PVD.
A PVD was confirmed on OCT if the posterior vitreous cortex or details of the vitreous body, bursa pre-macularis, hyperreflective vitreous strands, or granular details of the posterior vitreous body were absent from the retinal surface (Figure 2). In cases where scanned images were too dark to appreciate vitreous detail, contrast settings were modified to optimize such examination. These findings were recorded by two independent graders. If there was disagreement on the classification by the two graders on any of the OCT scans, the image was arbitrated by a third grader.
Outcomes
Treatment for DME was recorded during follow-up and included need for any intravitreous vascular endothelial growth factor (VEGF) inhibitor (aflibercept [Eylea; Regeneron, Tarrytown, NY], bevacizumab [Avastin; Genentech, South San Francisco, CA], or ranibizumab [Lucentis; Genentech, South San Francisco, CA]), intravitreous corticosteroid injection (triamcinolone, dexamethasone implant [Ozurdex; Allergan, Irvine, CA], or fluocinolone implant), or focal/grid photocoagulation. Need for treatment was determined based on the presence of center-involving DME with reduction in vision to 20/40 (Snellen) or worse. Other relevant information recorded included gender, race, lens status, hypertension, hyperlipidemia, insulin use, and type of diabetes to determine underlying prognostic factors. Hypertension, hyperlipidemia, and insulin use were confirmed by review of their medical chart at the initial visit.
Statistical Analysis
Status of PVD and need for DME therapy were evaluated among genders, ethnicities, and diabetes type using a Chi-squared test. Comparisons of age and HbA1c relative to PVD or therapy status were evaluated with a student's t-test. Cox proportional hazards model was used to evaluate the independent effects of PVD status, gender, ethnicity, age, HbA1c, baseline lens status, and diabetes type on the incidence of treatment for DME. A two-tailed P value of less than .05 was considered significant. SAS version 9.4 (2012; SAS Institute, Cary, NC) was used for data analysis.
Results
Data were collected for 689 eyes from 374 patients with diabetes and without PDR, DME, or prior DR/DME treatment. Eighty-three untreated eyes were removed from the dataset because follow-up was less than 2 years. Of those remaining eyes, 108 eyes had missing HbA1c data, two had unknown diabetes type, and two had unknown lens status. After removal of these eyes, data for 494 eyes from 274 patients were included in the final analysis. A total of 34 eyes (6.9%) were classified as having a complete PVD. Mean follow-up time for eyes with a PVD (1,470.9 days ± 553.0 days) was longer than eyes without a PVD (1,176.9 days ± 725.4 days; P = .004). On average, patients with PVD were older than patients without PVD (Table 1; 70.7 years ± 9.9 years vs. 60.8 years ± 11.5 years; P < .001). Complete PVD at baseline was also more frequent in non-African-Americans compared with African-Americans (P = .03). We did not observe any gender differences in relation to PVD status (P = .94).
A total of 146 eyes (29.6%) received treatment for DME during follow-up. Patients receiving therapy were younger than those not receiving therapy (Table 1; 59.1 years ± 11.5 years vs. 62.5 years ± 11.6 years; P = .003). The need for at least one DME treatment was more frequent in males versus females (P = .04). More patients on insulin therapy were treated for DME than patients not taking insulin (P = .01). Patients receiving treatment had higher HbA1c levels than those not requiring treatment (8.5 ± 2.4 vs. 7.8 ± 1.9; P < .001). Only 5.9% of eyes with a complete PVD required treatment during follow-up, whereas 31% of eyes without a complete PVD required treatment (Table 1; P = .001, Chi-squared).
We performed a multivariate analysis to determine any independent effects of PVD status on need for DME treatment. Cox proportional hazards modeling showed that after adjusting for age, ethnicity, gender, HbA1c, and diabetes type, a complete PVD was significantly associated with a reduction in hazard for DME therapy when compared to no PVD (hazard ratio [HR]: 0.18; 95% confidence interval [CI], 0.05–0.73; P = .02) (Table 2). All other classifications of incomplete PVD (Stage 1–3) were not protective (Table 2). In our cohort, female gender was independently associated with a decrease in hazard for DME therapy after controlling for other variables (HR: 0.68; 95% CI, 0.48–0.95; P = .02). Elevated HgbA1c was associated with an increase in hazard for DME therapy (HR: 1.11; 95% CI, 1.03–1.20; P = .005) (Table 2). Furthermore, therapeutic intervention was required earlier for patients with any baseline vitreomacular adhesions compared to those with a complete PVD, as visualized by a time-to-treatment analysis (Figure 3).
Discussion
In this study, multivariate analysis identified lower HbA1c, female gender, and complete PVD as factors associated with reduced treatment burden in DME. The association between elevated HbA1c levels and need for treatment in DME is well-established.13 Similarly, there is precedent for an association between female gender and reduced severity of DR and DME.14,15 In contrast, the effects of PVD on treatment need in DME are less well-understood. To investigate such effects, we classified the various stages of vitreous detachment (PVD Stage 0–4) using baseline SD-OCT findings (Figure 1). Our sub-analysis showed that any degree of vitreomacular adhesion at the baseline visit was associated with earlier and more frequent need for DME therapy (Table 2, Figure 3). A previous analysis conducted by the Diabetic Retinopathy Clinical Research Network (DRCR.net), showed that vitrectomy is beneficial in eyes with macular edema and VMT.16 Since eyes with overt VMT were excluded from this study, our findings are novel and reveal an important variable to consider in the management of diabetic eye disease.
Possible explanations for our findings are that any posterior vitreoretinal adhesions may serve as a conduit for retinal neovascularization or that these adhesions, no matter how small, may contribute to subclinical vitreoretinal traction. However, it is also possible that local oxygen tension increases sufficiently to suppress retinal VEGF secretion only in conditions where complete PVDs are present. Such a model would be consistent with data in vitrectomized eyes.5 Complete PVD may also account for the inverse relationship between degree of myopia and severity of DR that has been observed in numerous other studies, since myopic eyes are expected to have higher rates of complete PVD.17,18 A major limitation of this study is that it was not poised to address these contributions, since data on refractive errors were unavailable from both participating centers — both of which are tertiary academic referral centers for retinal disease. However, our findings are hypothesis-generating, suggesting an independent effect of PVD on need for DME treatment, and justify prospective analysis with more complete consideration of other confounding variable, such as myopia.
We could have used several methods to assess VMI status. Slit-lamp biomicroscopy, the most common method for inspecting the vitreous, has 80% sensitivity for detecting PVD in non-diabetic eyes.8 However, this method may be confounded in patients with diabetes due to the high prevalence of vitreoschisis, which is the splitting of the posterior vitreous during syneresis.19 In vitreoschisis, the posterior aspect of the vitreous remains attached to the retina, although more anterior layers separate away from the retina.20 Compared to the general population, eyes of patients with diabetes are more likely to exhibit both clinical and ultrastructural vitreoschisis and this may be caused by increases in glycation end products and hyperglycemia-induced crosslinking of collagen.1,21 On clinical examination, the anterior layer of the split vitreous may look similar to a shallow PVD even when a posterior layer of the vitreous cortex is still attached to the inner limiting membrane (Figures 2C and 2D).22 Like biomicroscopic methods, ultrasonographic assessments of PVD can be confounded by vitreoschisis. In comparison to these methods, SD-OCT is less prone to false positive errors given its greater resolution of vitreous details. Perhaps even more accurate will be the use of swept-source OCT (SS-OCT) coupled to widefield imaging, which will provide greater depths of imaging and the capability of viewing more of the posterior vitreomacular contact zone. Such technologies were not widely available at the outset of this study.
A limitation of using SD-OCT to assess for a complete PVD is that the detached posterior vitreous cortex may not be visible due to the narrow imaging window of modern tomography machines, relative to the vitreous cavity. Conversely, it is challenging to visualize the posterior vitreous on OCT when the vitreous is completely attached to the retina as it may be difficult to distinguish from the internal limiting membrane. However, several vitreous structures can be observed on SD-OCT as evidence of an attached vitreous, such as the bursa pre-macularis, posterior vitreous cortex, hyperreflective vitreous strands, and granular opacities.11,23,24 Although SD-OCT has its own limitations in detecting these structures, manipulation of the contrast and gain on the scan enhances their visualization (Figures 2C and 2D).23–25 Such techniques allowed us to make reliable PVD determinations in all cases included in this report. Several eyes diagnosed with a complete PVD on slit-lamp biomicroscopy were classified differently after visualization of posterior vitreous structures on OCT scan, indicating a completely attached vitreous (Figure 2). These examples illustrate the limitations of slit-lamp biomicroscopy in evaluating PVD.
There are caveats to this retrospective study. First, patients who initially presented with incomplete PVD may have developed a PVD over the course of our observation period, which was — on average — 39.4 months. Despite this caveat, initial PVD status is independently associated with DME treatment by multivariable analysis, and this finding warrants further analysis in a prospective format. Second, the limited view in a standard macular cube scan (6 mm × 6 mm) compared to a widened view (12 mm × 9 mm) with SS-OCT made the evaluation of PVD more difficult. However, our use of independent graders minimized errors in grading, and all OCT scans included in this study were successfully graded. Third, not all potential variables contributing to treatment need in DME were measured, namely refractive status. Myopia is a protective factor in the development of DR and DME, and therefore could have influenced rate of treatment between groups since complete PVD is more likely in myopic eyes,17,18 but we were unable to assess the potential contributions of refractive errors in relation to PVD status and DME treatment since these data were not available in our cohort. Lastly, the relatively small sample size and higher mean age of patients with complete PVD may have affected outcomes. However, in a multivariate Cox regression analysis taking into account both the small sample of patients with complete PVD and mean age difference between groups, PVD status was an independent predictor of need for treatment whereas age was not (Table 2).
The strengths of this study include its large sample size of 494 eyes and its use of a sensitive, relatively unbiased method to grade vitreoretinal interface status. Our findings suggest that a complete PVD is protective in terms of need for treatment in patients with DME. With technological advancements, including SS-OCT, better visualization of VMI details may reveal more about the relationship of the vitreous to diabetic eye disease and provide more precise prognostic information for those suffering from it.
References
- Nesmith BL, Palacio AC, Schaal Y, Gupta A, Schaal S. Diabetes Alters the Magnitude of Vitreomacular Adhesion. Retina. 2017;37(4):749–752. https://doi.org/10.1097/IAE.0000000000001228 PMID: doi:10.1097/IAE.0000000000001228 [CrossRef]
- Agarwal D, Gelman R, Prospero Ponce C, Stevenson W, Christoforidis JB. The Vitreomacular Interface in Diabetic Retinopathy. J Ophthalmol. 2015;2015:392983. https://doi.org/10.1155/2015/392983 PMID:26425349
- Ono R, Kakehashi A, Yamagami H, et al. Prospective assessment of proliferative diabetic retinopathy with observations of posterior vitreous detachment. Int Ophthalmol. 2005;26(1–2):15–19. https://doi.org/10.1007/s10792-005-5389-2 PMID: doi:10.1007/s10792-005-5389-2 [CrossRef]
- Akiba J, Arzabe CW, Trempe CL. Posterior vitreous detachment and neovascularization in diabetic retinopathy. Ophthalmology. 1990;97(7):889–891. https://doi.org/10.1016/S0161-6420(90)32486-7 PMID: doi:10.1016/S0161-6420(90)32486-7 [CrossRef]2381702
- Holekamp NM, Shui YB, Beebe DC. Vitrectomy surgery increases oxygen exposure to the lens: a possible mechanism for nuclear cataract formation. Am J Ophthalmol. 2005;139(2):302–310. https://doi.org/10.1016/j.ajo.2004.09.046 PMID: doi:10.1016/j.ajo.2004.09.046 [CrossRef]15733992
- Kado M, Jalkh AE, Yoshida A, et al. Vitreous changes and macular edema in central retinal vein occlusion. Ophthalmic Surg. 1990;21(8):544–549. PMID:2234801
- Gella L, Raman R, Kulothungan V, Sharma T. Prevalence of posterior vitreous detachment in the population with type II diabetes mellitus and its effect on diabetic retinopathy: Sankara Nethralaya Diabetic Retinopathy Epidemiology and Molecular Genetic Study SN-DREAMS report no. 23. Jpn J Ophthalmol. 2012;56(3):262–267. https://doi.org/10.1007/s10384-012-0134-7 PMID: doi:10.1007/s10384-012-0134-7 [CrossRef]22476626
- Kičová N, Bertelmann T, Irle S, Sekundo W, Mennel S. Evaluation of a posterior vitreous detachment: a comparison of biomicroscopy, B-scan ultrasonography and optical coherence tomography to surgical findings with chromodissection. Acta Ophthalmol. 2012;90(4):e264–e268. https://doi.org/10.1111/j.1755-3768.2011.02330.x PMID: doi:10.1111/j.1755-3768.2011.02330.x [CrossRef]
- Gaucher D, Tadayoni R, Erginay A, Haouchine B, Gaudric A, Massin P. Optical coherence tomography assessment of the vitreoretinal relationship in diabetic macular edema. Am J Ophthalmol. 2005;139(5):807–813. https://doi.org/10.1016/j.ajo.2004.12.084 PMID: doi:10.1016/j.ajo.2004.12.084 [CrossRef]15860284
- Itakura H, Kishi S. Evolution of vitreomacular detachment in healthy subjects. JAMA Ophthalmol. 2013;131(10):1348–1352. https://doi.org/10.1001/jamaophthalmol.2013.4578 PMID: doi:10.1001/jamaophthalmol.2013.4578 [CrossRef]23974841
- Mojana F, Kozak I, Oster SF, et al. Observations by spectral-domain optical coherence tomography combined with simultaneous scanning laser ophthalmoscopy: imaging of the vitreous. Am J Ophthalmol. 2010;149(4):641–650. https://doi.org/10.1016/j.ajo.2009.11.016 PMID: doi:10.1016/j.ajo.2009.11.016 [CrossRef]20138610
- Johnson MW. Perifoveal vitreous detachment and its macular complications. Trans Am Ophthalmol Soc. 2005;103:537–567. PMID:17057817
- The relationship of glycemic exposure (HbA1c) to the risk of development and progression of retinopathy in the diabetes control and complications trial. Diabetes. 1995;44(8):968–983. https://doi.org/10.2337/diab.44.8.968 PMID: doi:10.2337/diab.44.8.968 [CrossRef]7622004
- Ozawa GY, Bearse MA Jr, Adams AJ. Male-female differences in diabetic retinopathy?Curr Eye Res. 2015;40(2):234–246. https://doi.org/10.3109/02713683.2014.958500 PMID: doi:10.3109/02713683.2014.958500 [CrossRef]
- Stratton IM, Kohner EM, Aldington SJ, et al. UKPDS 50: risk factors for incidence and progression of retinopathy in Type II diabetes over 6 years from diagnosis. Diabetologia. 2001;44(2):156–163. https://doi.org/10.1007/s001250051594 PMID: doi:10.1007/s001250051594 [CrossRef]11270671
- Haller JA, Qin H, Diabetic Retinopathy Clinical Research Network Writing Committee, et al. Vitrectomy outcomes in eyes with diabetic macular edema and vitreomacular traction. Ophthalmology. 2010;117(6):1087–1093.e3. https://doi.org/10.1016/j.ophtha.2009.10.040 PMID: doi:10.1016/j.ophtha.2009.10.040 [CrossRef]20299105
- Bazzazi N, Akbarzadeh S, Yavarikia M, Poorolajal J, Fouladi DF. HIGH MYOPIA AND DIABETIC RETINOPATHY: A Contralateral Eye Study in Diabetic Patients With High Myopic Anisometropia. Retina. 2017;37(7):1270–1276. https://doi.org/10.1097/IAE.0000000000001335 PMID: doi:10.1097/IAE.0000000000001335 [CrossRef]
- Fu Y, Geng D, Liu H, Che H. Myopia and/or longer axial length are protective against diabetic retinopathy: a meta-analysis. Acta Ophthalmol. 2016;94(4):346–352. https://doi.org/10.1111/aos.12908 PMID: doi:10.1111/aos.12908 [CrossRef]
- Sebag J. Diabetic vitreopathy. Ophthalmology. 1996;103(2):205–206. https://doi.org/10.1016/S0161-6420(96)30716-1 PMID: doi:10.1016/S0161-6420(96)30716-1 [CrossRef]8594502
- Sebag J. Vitreoschisis SJ. Graefes Arch Clin Exp Ophthalmol. 2008;246(3):329–332. https://doi.org/10.1007/s00417-007-0743-x PMID: doi:10.1007/s00417-007-0743-x [CrossRef]18228032
- Gandorfer A, Rohleder M, Grosselfinger S, Haritoglou C, Ulbig M, Kampik A. Epiretinal pathology of diffuse diabetic macular edema associated with vitreomacular traction. Am J Ophthalmol. 2005;139(4):638–652. https://doi.org/10.1016/j.ajo.2004.11.035 PMID: doi:10.1016/j.ajo.2004.11.035 [CrossRef]15808159
- Kakehashi A, Takezawa M, Akiba J. Classification of posterior vitreous detachment. Clin Ophthalmol. 2014;8:1–10. PMID:24376338
- Liu JJ, Witkin AJ, Adhi M, et al. Enhanced vitreous imaging in healthy eyes using swept source optical coherence tomography. PLoS One. 2014;9(7):e102950. https://doi.org/10.1371/journal.pone.0102950 PMID: doi:10.1371/journal.pone.0102950 [CrossRef]25036044
- Shimada H, Hirose T, Yamamoto A, Nakashizuka H, Hattori T, Yuzawa M. Depiction of the vitreous pocket by optical coherence tomography. Int Ophthalmol. 2011;31(1):51–53. https://doi.org/10.1007/s10792-010-9398-4 PMID: doi:10.1007/s10792-010-9398-4 [CrossRef]
- Itakura H, Kishi S. Aging changes of vitreomacular interface. Retina. 2011;31(7):1400–1404. https://doi.org/10.1097/IAE.0b013e318206cb43 PMID: doi:10.1097/IAE.0b013e318206cb43 [CrossRef]21233785
Cohort Demographics
| PVD n (%) | No PVD n (%) | P Value | Therapy n (%) | No Therapy n (%) | P Value |
---|
|
---|
Eyes | 34 (6.9) | 460 (93.1) | | 146 (29.6) | 348 (70.4) | |
|
Gender | | | .94 | | | .04 |
Male | 13 (38.2) | 173 (37.6) | | 65 (44.5) | 121 (34.8) | |
Female | 21 (61.8) | 287 (62.4) | | 81 (55.5) | 227 (65.2) | |
|
Mean age (years) | 70.7 ± 9.9 | 60.8 ± 11.5 | < .001 | 59.1 ± 11.5 | 62.5 ± 11.6 | .003 |
|
Diabetes Type | | | .05 | | | .41 |
1 | 0 (0) | 46 (10.0) | | 16 (11.0) | 30 (8.6) | |
2 | 34 (100.0) | 414 (90.0) | | 130 (89.0) | 318 (91.4) | |
|
HbA1c | 7.4 ± 1.9 | 8.0 ± 2.1 | .08 | 8.5 ± 2.4 | 7.8 ± 1.9 | < .001 |
|
Ethnicity | | | .02 | | | .57 |
Caucasian and Other | 23 (67.6) | 221 (48.0) | | 75 (51.4) | 169 (48.6) | |
African American | 11 (32.4) | 239 (52.0) | | 71 (48.6) | 179 (51.4) | |
|
Lens Status | | | .04 | | | .06 |
Phakic | 22 (64.7) | 364 (79.1) | | 122 (83.6) | 264 (75.9) | |
Pseudophakic | 12 (35.3) | 96 (20.9) | | 24 (16.4) | 84 (24.1) | |
|
Hypertension | | | .61 | | | .18 |
Yes | 19 (100) | 305 (93.8) | | 79 (97.5) | 245 (93.2) | |
No | 0 | 20 (6.2) | | 2 (2.5) | 18 (6.8) | |
|
Hyperlipidemia | | | .18 | | | .88 |
Yes | 17 (6.6) | 240 (73.8) | | 60 (74.1) | 197 (74.9) | |
No | 2 (2.3) | 85 (26.2) | | 21 (25.9) | 66 (25.1) | |
|
Insulin Use | | | .05 | | | .01 |
Yes | 9 (47.4) | 223 (68.6) | | 64 (79.0) | 168 (63.9) | |
No | 10 (52.6) | 102 (31.4) | | 17 (21.0) | 95 (36.1) | |
|
DME Treatment | | | .001 | | | |
No Treatment | 32 (94.1) | 316 (90.8) | | | | |
≥ 1 Therapy | 2 (5.9) | 32 (9.2) | | | | |
Multivariate Cox Regression Analyses on Probability of Intervention for DME Therapy
| Adjusted HR (95% CI) | P Value |
---|
|
---|
PVD Stage | | |
0 (n = 299) | Ref | |
1 (n = 112) | 0.86 (0.57, 1.30) | .48 |
2 (n = 18) | 2.00 (0.90, 4.44) | .09 |
3 (n = 31) | 1.05 (0.52, 2.12) | .90 |
4 (n = 34) | 0.18 (0.05, 0.73) | .02 |
|
Lens Status | | |
Phakic (n = 386) | Ref | |
Pseudophakic (n = 108) | 0.70 (0.44, 1.12) | .14 |
|
Diabetes Type | | |
1 (n = 46) | 0.89 (0.40, 1.98) | .78 |
2 (n = 448) | Ref | |
|
HbA1c | 1.11 (1.03, 1.20) | .005 |
|
Gender | | |
Male (n = 308) | Ref | |
Female (n = 186) | 0.68 (0.48, 0.95) | .02 |
|
Age | | |
≤39 (n = 22) | Ref | |
40–49 (n = 37) | 0.78 (0.28, 2.22) | .64 |
50–59 (n = 141) | 0.59 (0.23, 1.48) | .26 |
60–69 (n = 179) | 0.76 (0.29, 2.01) | .58 |
≥70 (n = 115) | 0.50 (0.17, 1.45) | .20 |
|
Ethnicity | | |
Caucasian and other (n = 244) | Ref | |
African-American (n = 250) | 0.89 (0.62, 1.28) | .54 |