Marginal Regression Models Research Articles

Marginal regression models with time‐varying coefficients for recurrent event data

Recurrent event data arise frequently from medical research. Examples include repeated infections, recurrence of tumors, relapse of leukemia, repeated hospitalizations, recurrence of symptoms of a disease, and so on. In the analysis of recurrent event data, the proportional rates model assumes that the regression coefficients are time invariant. In reality, however, these parameters may vary over time, and the temporal covariate effects on the event process are of great interest. In this article, we formulate a class of semiparametric marginal rates models, which incorporate a mixture of time-varying and time-independent parameters, to analyze recurrent event data. For statistical inference on model parameters, an estimation procedure is developed and asymptotic properties of the proposed estimators are established. In addition, we develop tests for investigating whether or not covariate effects vary with time. The finite-sample behaviors of the proposed methods are examined in simulation studies. An example of application of the proposed methodology is illustrated on a set of data from a clinic study on chronic granulomatous disease.

Soluble triggering receptor expressed on myeloid cells-1 as an infection marker for patients with neutropenic fever*

To assess the value of soluble triggering receptor expressed on myeloid cells-1 as a biomarker of infection in patients with neutropenic fever comparing with procalcitonin and C-reactive protein. Prospective, comparative, single-center study. Hematology ward at a university hospital. Seventy-five patients with neutropenic fever after chemotherapy for their hematologic malignancies. None. A total of 137 episodes of neutropenic fever in 75 patients were classified into 75 episodes of documented infections and 62 low likelihood of infection. The level of soluble triggering receptor expressed on myeloid cells-1 was significantly elevated in the group of documented infection. The area under the receiver operating characteristic curve for the diagnosis of infection was 0.719 (95% confidence interval, 0.632-0.806; p < .0001) for soluble triggering receptor expressed on myeloid cells-1, which was larger than the values of 0.501 for procalcitonin (0.465-0.657; p = .218) and 0.491 for C-reactive protein (0.394-0.589, p = .858). The fitted marginal logistic regression model of all episodes contained two statistically significant predictors of infection: soluble triggering receptor expressed on myeloid cells-1 (per 1-pg/mL increase; odds ratio [OR], 1.0002; 95% CI, 1.0001-1.0003; p < .0001) and procalcitonin (per 1-ng/mL increase; OR, 1.0094; 95% CI, 1.0005-1.0184; p = .0002). In a diagnostic panel with soluble triggering receptor expressed on myeloid cells-1 and procalcitonin, the sensitivity and specificity were 88% and 48%, respectively. Soluble triggering receptor expressed on myeloid cells-1 is better than procalcitonin in the prediction of infection at the onset of neutropenic fever. By applying soluble triggering receptor expressed on myeloid cells-1 and procalcitonin together, low or high risk for infection can be defined at the onset of neutropenic fever.

Influence of Disc Size on Optic Nerve Head versus Retinal Nerve Fiber Layer Assessment for Diagnosing Glaucoma

To explore and compare the influence of optic disc size on the diagnostic accuracy of retinal nerve fiber layer (RNFL) thickness and optic nerve head (ONH) quantitative assessment. Observational, cross-sectional evaluation of diagnostic tests. We included 120 eyes from 50 normal subjects and 70 glaucomatous patients classified by the presence of a repeatable visual field defect for the analysis. The RNFL thickness was measured by scanning laser polarimetry with variable corneal compensator (GDx-VCC, Carl-Zeiss Meditec, Dublin, CA) and spectral-domain optical coherence tomography (Cirrus HD-OCT, Carl Zeiss Meditec, Inc). We obtained ONH imaging by means of confocal scanning laser ophthalmoscopy (HRT3; Heidelberg Engineering, GmbH, Dossenheim, Germany). Sensitivity and specificity for normative classifications, sensitivity at fixed specificity and area under the receiver operating characteristics curve (AUC) for continuous parameters. A logistic marginal regression model and coefficients of variation (CoV) have been used to test and quantify the influence of optic disc size on the diagnostic accuracy of the 3 technologies under investigation. Among continuous parameters average RNFL thickness for Cirrus HD-OCT, nerve fiber indicator for GDx-VCC and cup shape measure for the HRT3 showed the best diagnostic accuracy with an AUC of 0.97, 0.94, and 0.94, respectively. Among normative classifications, the highest sensitivity and specificity were found for OCT average RNFL thickness (75.8% and 94.7%), for GDx superior thickness (77.1% and 97.5%), for HRT3 Moorfields regression analysis result (89.4% and 73.7%) and for HRT3 GPS global (92.3% and 76.5%). The diagnostic performance of HRT3 parameters seemed to be significantly influenced by optic disc size, although the same was not true for Cirrus HD-OCT and GDx VCC. The most steady performers for each imaging device across disc size groups were Cirrus HD-OCT average thickness (CoV, 1.6%), GDx-VCC inferior thickness (CoV, 2.5%), and HRT3 GPS temporal and nasal (CoV, 21.4%). The diagnostic accuracy of quantitative RNFL assessment as performed by Cirrus HD-OCT and GDx-VCC is high and virtually unaffected or only minimally affected by the size of the optic disc and may provide more consistent diagnostic outcomes across small and large discs than ONH assessment as performed by HRT3.

Regression Analysis for Recurrent Events Data under Dependent Censoring

Recurrent events data are commonly seen in longitudinal follow-up studies. Dependent censoring often occurs due to death or exclusion from the study related to the disease process. In this article, we assume flexible marginal regression models on the recurrence process and the dependent censoring time without specifying their dependence structure. The proposed model generalizes the approach by Ghosh and Lin (2003, Biometrics 59, 877-885). The technique of artificial censoring provides a way to maintain the homogeneity of the hypothetical error variables under dependent censoring. Here we propose to apply this technique to two Gehan-type statistics. One considers only order information for pairs whereas the other utilizes additional information of observed censoring times available for recurrence data. A model-checking procedure is also proposed to assess the adequacy of the fitted model. The proposed estimators have good asymptotic properties. Their finite-sample performances are examined via simulations. Finally, the proposed methods are applied to analyze the AIDS linked to the intravenous experiences cohort data.

Copula‐based regression models for a bivariate mixed discrete and continuous outcome

This paper is concerned with regression models for correlated mixed discrete and continuous outcomes constructed using copulas. Our approach entails specifying marginal regression models for the outcomes, and combining them via a copula to form a joint model. Specifically, we propose marginal regression models (e.g. generalized linear models) to link the outcomes' marginal means to covariates. To account for associations between outcomes, we adopt the Gaussian copula to indirectly specify their joint distributions. Our approach has two advantages over current methods: one, regression parameters in models for both outcomes are marginally meaningful, and two, the association is 'margin-free', in the sense that it is characterized by the copula alone. By assuming a latent variable framework to describe discrete outcomes, the copula used still uniquely determines the joint distribution. In addition, association measures between outcomes can be interpreted in the usual way. We report results of simulations concerning the bias and efficiency of two likelihood-based estimation methods for the model. Finally, we illustrate the model using data on burn injuries.

Estimating functions for evaluating treatment effects in cluster‐randomized longitudinal studies in the presence of drop‐out and non‐compliance

AbstractWe describe methods for analyzing longitudinal binary data from cluster‐randomized trials in which responses are incompletely observed and subjects may not be fully compliant with the prescribed treatment regimen. The method is based on a marginal regression model for the response where parameter estimates are obtained from generalized estimating equations. Estimating equations are also employed to estimate parameters of the missing data process which are used to compute inverse probability weights. A model is specified for the compliance process which facilitates estimating the expectation of the contributions to the estimating function for the response parameters among individuals without compliance data, which occurs when the control treatment involves no intervention. The approach is robust in the sense that semi‐parametric models are used for the response and the missing data processes and robust variance estimates are advocated. The proposed method is shown to perform well in simulation studies, and data from a randomized trial of patients with depression are analyzed for illustration. The Canadian Journal of Statistics 38: 232–255; 2010 © 2010 Statistical Society of Canada

Diagnostic Ability of Heidelberg Retina Tomography in Detecting Glaucoma in a Population Setting: The Singapore Malay Eye Study

To evaluate the performance of the Heidelberg Retina Tomograph II (HRT II, Heidelberg Engineering, Heidelberg, Germany) in diagnosing glaucoma in a population setting. Population-based cross-sectional study. Of 3280 Malay persons aged 40 to 80 years who participated in the survey, 112 subjects (124 eyes) with glaucoma and a subset of 196 bilaterally normal subjects (392 eyes) were included for the evaluation of diagnostic ability of HRT II. All glaucomatous and normal eyes underwent standardized ophthalmic assessment (including automated perimetry) and HRT II. Glaucoma was defined according to International Society for Geographical and Epidemiological Ophthalmology criteria. Area under the curve (AUC) receiver operating characteristic (ROC), sensitivity, and specificity were used to evaluate the diagnostic performance of HRT II algorithms. Marginal logistic regression models were used to evaluate the influence of optic disc size on the results of HRT II algorithms. The HRT II algorithms: linear discriminant functions (LDFs) by Mikelberg et al (LDF1), Burk et al (LDF2), Bathija et al (LDF3), and Moorfields regression analysis (MRA). The MRA was subdivided into MRA1 with "borderline" outcomes as positive and MRA2 with "borderline" outcomes as negative. Subjects with cataract, visual impairment, astigmatism, and greater negative spherical equivalent, and of older age were more likely to yield lower quality images. For analyses by eye, AUCs were 0.789, 0.704, 0.755, 0.754, and 0.762 for MRA1, MRA2, LDF1, LDF2, and LDF3, respectively. At 85% specificity, sensitivities were 62.1%, 65.3%, and 66.9% for LDF1, LDF2, and LDF3, respectively. At 95% specificity, these figures decreased to 31.5%, 42.7%, and 45.2%, respectively. The sensitivity and specificity were 71.0% and 86.7% for MRA1 and 43.6% and 97.2% for MRA2, respectively. Similar estimates were found for analyses by person. Larger optic disc size was associated with increased sensitivity and false-positive rate for MRA1, LDF1, and LDF2. LDF1 and LDF3 were least affected by optic disc area, but the sensitivities were moderate and the false-positive rates were high across different optic disc areas. The current HRT II algorithms are of limited value for population-based glaucoma screening in the Malay population and do not account adequately for optic disc size.

Semiparametric Efficient Estimation for Incomplete Longitudinal Binary Data, With Application to Smoking Trends

Incomplete longitudinal data often are analyzed with estimating equations for inference on a parameter from a marginal mean regression model. Generalized estimating equations, although commonly used for incomplete longitudinal data, are invalid for data that are not missing completely at random. There exists a class of inverse probability weighted estimating equations that are valid under dropouts missing at random, including an easy-to-implement but inefficient member. A relatively computationally complex semiparametric efficient estimator in this class has been applied to continuous data. A specific form of this estimator is developed for binary data and used as a benchmark for assessing the efficiency of the simpler estimator in a simulation study. Both are applied in the estimation of 15-year cigarette smoking trends in the United States from a cohort of 5077 young adults. The results suggest that declines in smoking from previous reports have been exaggerated.

Marginal hazards model for case-cohort studies with multiple disease outcomes

Case-cohort study designs are widely used to reduce the cost of large cohort studies while achieving the same goals, especially when the disease rate is low. A key advantage of the case-cohort study design is its capacity to use the same subcohort for several diseases or for several subtypes of disease. In order to compare the effect of a risk factor on different types of diseases, times to different events need to be modelled simultaneously. Valid statistical methods that take the correlations among the outcomes from the same subject into account need to be developed. To this end, we consider marginal proportional hazards regression models for case-cohort studies with multiple disease outcomes. We also consider generalized case-cohort designs that do not require sampling all the cases, which is more realistic for multiple disease outcomes. We propose an estimating equation approach for parameter estimation with two different types of weights. Consistency and asymptotic normality of the proposed estimators are established. Large sample approximation works well in small samples in simulation studies. The proposed methods are applied to the Busselton Health Study.

Impact of multiple caregiving roles on elevated depressed mood in early-stage breast cancer patients and same-age controls

The effect of caregiving roles on risk of elevated depressed mood over 12 months was examined in early-stage (0-IIA) breast cancer patients and same-aged women without breast cancer. Women were interviewed 4-6 weeks, 6 months, and 12 months following definitive surgical treatment (patients) or routine screening mammogram (controls). The Center for Epidemiologic Studies-Depression Scale was administered at each interview and dichotomized for analysis (<16 [little/no depressed mood] vs. >or=16 [elevated depressed mood]). Participants were categorized as having no caregiving responsibilities, caregiving for children or other persons, or caregiving for both children and others (multiple caregiving roles). Two multivariable marginal logistic regression models with repeated measures were fit (one each for patients and controls) to examine the effect of caregiving roles on elevated depressed mood, using generalized estimating equations to account for intra-individual correlations. Of 1096 participants (mean age 58; 76% white), 1019 with caregiving data were included in the analysis. Compared with baseline, patients with multiple caregiving roles (23/521 patients) were at increased risk of elevated depressed mood at 6 months (adjusted odds ratio [aOR], 7.20; 95% confidence interval [CI], 1.17-44.46; P = 0.034), and controls with multiple caregiving roles (15/498 controls) were at decreased risk of elevated depressed mood at 12-month follow-up (aOR, 0.12; 95% CI, 0.02-0.97; P = 0.047). Patients with multiple caregiving roles were more likely while controls were less likely to report elevated depressed mood over time, suggesting a need to identify patients with multiple caregiving roles early during their treatment.

Pre-radiographic MRI findings are associated with onset of knee symptoms: the most study

Magnetic resonance imaging (MRI) has greater sensitivity to detect osteoarthritis (OA) damage than radiographs but it is uncertain which MRI findings in early OA are clinically important. We examined MRI abnormalities detected in knees without radiographic OA and their association with incident knee symptoms. Participants from the Multicenter Osteoarthritis Study (MOST) without frequent knee symptoms (FKS) at baseline were eligible if they also lacked radiographic features of OA at baseline. At 15 months, knees that developed FKS were defined as cases while control knees were drawn from those that remained without FKS. Baseline MRIs were scored at each subregion for cartilage lesions (CARTs); osteophytes (OST); bone marrow lesions (BML) and cysts. We compared cases and controls using marginal logistic regression models, adjusting for age, gender, race, body mass index (BMI), previous injury and clinic site. 36 case knees and 128 control knees were analyzed. MRI damage was common in both cases and controls. The presence of a severe CART (P=0.03), BML (P=0.02) or OST (P=0.02) in the whole knee joint was more common in cases while subchondral cysts did not differ significantly between cases and controls (P>0.1). Case status at 15 months was predicted by baseline damage at only two locations; a BML in the lateral patella (P=0.047) and at the tibial subspinous subregions (P=0.01). In knees without significant symptoms or radiographic features of OA, MRI lesions of OA in only a few specific locations preceded onset of clinical symptoms and suggest that changes in bone play a role in the early development of knee pain. Confirmation of these findings in other prospective studies of knee OA is warranted.

Optimal combination of estimating equations in the analysis of multilevel nested correlated data

Multilevel nested, correlated data often arise in biomedical research. Examples include teeth nested within quadrants in a mouth or students nested within classrooms in schools. In some settings, cluster sizes may be large relative to the number of independent clusters and the degree of correlation may vary across clusters. When cluster sizes are large, fitting marginal regression models using Generalized Estimating Equations with flexible correlation structures that reflect the nested structure may fail to converge and result in unstable covariance estimates. Also, the use of patterned, nested working correlation structures may not be efficient when correlation varies across clusters. This paper describes a flexible marginal regression modeling approach based on an optimal combination of estimating equations. Particular within-cluster and between-cluster data contrasts are used without specification of the working covariance structure and without estimation of covariance parameters. The method involves estimation of the covariance matrix only for the vector of component estimating equations (which is typically of small dimension) rather than the covariance matrix of the observations within a cluster (which may be of large dimension). In settings where the number of clusters is large relative to the cluster size, the method is stable and is highly efficient, while maintaining appropriate coverage levels. Performance of the method is investigated with simulation studies and an application to a periodontal study.

Quality of life in advanced non-small cell lung cancer patients receiving first-line gefitinib monotherapy

9614 Background: Gefitinib is a potential first-line treatment option for patients with advanced non-small cell lung cancer (NSCLC), especially for patients with activating mutations in the EGFR gene. However, little is known about patient-reported health-related quality of life (HRQOL) in this patient population. The aims of this study were to explore the prognostic values of baseline HRQOL for time-to-treatment failure (TTF), as well as the predictors of repeatedly measured posttreatment HRQOL, in advanced NSCLC patients receiving first-line gefitinib. Methods: A total of 106 chemonaive patients with advanced NSCLC were enrolled in a phase II trial. Gefitinib was given at a dose of 250 mg/d. HRQOL was assessed monthly with the EuroQoL instrument (EQ-5D) and the Lung Cancer Symptom Scale (LCSS) questionnaire. Baseline HRQOL and clinical/molecular predictors of TTF were jointly examined by multiple Cox's proportional hazards model. The associations between the clinical/molecular factors and repeatedly measured posttreatment HRQOL were analyzed by fitting marginal linear regression model using the generalized estimating equations (GEE) method. Results: In this prospective study, HRQOL data were obtained from 94 patients. Baseline EQ-5D index (estimated hazard ratio = 0.286, 95% C.I.: 0.135–0.603, p = 0.001) and the presence of L858R EGFR mutation in adenocarcinoma (estimated hazard ratio = 0.520, 95% C.I.: 0.307–0.880, p = 0.015) were retained as independent prognostic factors in the final multiple Cox's proportional hazards model for TTF. According to preliminary GEE analysis of repeatedly measured posttreatment HRQOL, the patients with wild-type EGFR consistently had worse HRQOL in EQ-5D index (p &lt; 0.0001), EQ-5D VAS score (p = 0.0002), and LCSS global score (p &lt; 0.0001), respectively. Conclusions: In advanced NSCLC patients receiving first-line gefitinib, better baseline EQ-5D index and L858R EGFR mutation in adenocarcinoma predict longer TTF. In addition, patients with wild-type EGFR had worse posttreatment HRQOL. No significant financial relationships to disclose.

Marginal Regression Model with Time-Varying Coefficients for Panel Data

In this article, we formulate a class of semiparametric marginal means models with a mixture of time-varying and time-independent parameters for analyzing panel data. For inference about the regression parameters, an estimation procedure is developed and asymptotic properties of the proposed estimators are established. In addition, some tests are presented for investigating whether or not covariate effects vary with time. The finite-sample behavior of the proposed methods is examined in simulation studies, and the data from an AIDS clinical trial study are used to illustrate the methodology.

Efficient parameter estimation in longitudinal data analysis using a hybrid GEE method

The method of generalized estimating equations (GEEs) provides consistent estimates of the regression parameters in a marginal regression model for longitudinal data, even when the working correlation model is misspecified (Liang and Zeger, 1986). However, the efficiency of a GEE estimate can be seriously affected by the choice of the working correlation model. This study addresses this problem by proposing a hybrid method that combines multiple GEEs based on different working correlation models, using the empirical likelihood method (Qin and Lawless, 1994). Analyses show that this hybrid method is more efficient than a GEE using a misspecified working correlation model. Furthermore, if one of the working correlation structures correctly models the within-subject correlations, then this hybrid method provides the most efficient parameter estimates. In simulations, the hybrid method's finite-sample performance is superior to a GEE under any of the commonly used working correlation models and is almost fully efficient in all scenarios studied. The hybrid method is illustrated using data from a longitudinal study of the respiratory infection rates in 275 Indonesian children.

Improving the Diagnostic Accuracy of Depression in Older Persons: The Depression in the Aged Female National Evaluation Cluster Randomized Trial

To evaluate whether a training intervention can improve the ability of geriatricians to recognize depression in older persons. Multicenter, cluster randomized clinical trial. Fourteen geriatric outpatient clinics in Italy, each representing the unit of randomization. After training, a total of 1,914 outpatients aged 65 years and older in both arms, not on antidepressant at entry, were blindly evaluated by the clinic geriatrician, in charge of routine clinical management, and by a field researcher, who formally diagnosed depression by the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), taken as the criterion standard. After randomization, geriatricians belonging to the intervention arm were assigned to receive a residential 3-day educational program on depression. Those in the control arm received a generic course on disease management in elderly people. Sensitivity and specificity of the diagnosis of depression made by geriatricians, compared with the DSM-IV diagnosis. Sensitivity and specificity were significantly higher in trained than in untrained geriatricians (49 vs 35% and 91 vs 88%, respectively; P=.002 in marginal regression models). Effectiveness of training was confirmed, adjusting for age, sex, and cognitive performance (P=.02). The ability of geriatricians to diagnose depression in older outpatients can be improved with a specific training intervention. Improvement of diagnostic performance might translate into more-appropriate clinical management.

Are Comparisons of Consumer Satisfaction With Providers Biased by Nonresponse or Case-Mix Differences?

Are Comparisons of Consumer Satisfaction With Providers Biased by Nonresponse or Case-Mix Differences?

A Screwless and Cementless Technique for the Restoration of Single‐tooth Implants: A Retrospective Cohort Study

The Integrated Abutment Crown (IAC) is a technique for the fabrication of single-tooth implant-supported crowns where the abutment and the crown are one unit. The abutment-crown complex is connected to the implant with a locking taper. This technique does not use cement to retain the crown or screws to retain the abutment. The purpose of this study was to evaluate the clinical outcome of screwless, cementless single implant-supported crowns (IACs) placed in a general dental practice. A retrospective cohort study was conducted between July 2001 and August 2003. Patients were recalled between January and March 2004. The restorations were evaluated following the modified United States Public Health Service (USPHS) criteria. Several other variables, such as anatomic form, occlusion, soft tissue health, and reconstructive procedures, were also recorded. Descriptive statistics, univariate and multivariate marginal Cox Proportional Hazards Regression models, adjusted for multiple implants in the same patient, were used. During the chart review, 108 patients were identified. A cohort of 59 patients with a total of 151 IACs met the inclusion criteria. The Kaplan-Meier survival rate for IACs was 98.7%. Two IACs were removed, one due to implant failure; the other became loose several times and was replaced with a splinted restoration. Excellent marginal adaptation was observed with no clinically discernible interface between the veneer material and the abutment. Nine maxillary anterior IACs loosened on five patients; eight of them were reinserted and continued in function without further problems for the remainder of the study. An IAC located between a tooth and an implant was 2.65 times more likely to have postinsertion complications (p= 0.05). An IAC with incorrect anatomic form (overcontoured) was 3.26 times more likely to have postinsertion complications (p= 0.01). Maxillary anterior IACs adjacent to one tooth and one implant were 3.9 times more likely to come loose (p= 0.05). The clinical outcome of this screwless and cementless system for single implant restorations compares favorably with the experience of screw- and cement-retained single implant restorations within the observation period.

Locally Dependent Latent Class Models with Covariates: An Application to Under-Age Drinking in the USA

Under-age drinking is a long-standing public health problem in the USA and the identification of underage drinkers suffering alcohol-related problems has been difficult by using diagnostic criteria that were developed in adult populations. For this reason, it is important to characterize patterns of drinking in adolescents that are associated with alcohol-related problems. Latent class analysis is a statistical technique for explaining heterogeneity in individual response patterns in terms of a smaller number of classes. However, the latent class analysis assumption of local independence may not be appropriate when examining behavioural profiles and could have implications for statistical inference. In addition, if covariates are included in the model, non-differential measurement is also assumed. We propose a flexible set of models for local dependence and differential measurement that use easily interpretable odds ratio parameterizations while simultaneously fitting a marginal regression model for the latent class prevalences. Estimation is based on solving a set of second-order estimating equations. This approach requires only specification of the first two moments and allows for the choice of simple 'working' covariance structures. The method is illustrated by using data from a large-scale survey of under-age drinking. This new approach indicates the effectiveness of introducing local dependence and differential measurement into latent class models for selecting substantively interpretable models over more complex models that are deemed empirically superior.

Scanning Laser Polarimetry with Variable Corneal Compensation in Primary Angle-closure Glaucoma

To evaluate the diagnostic sensitivity of scanning laser polarimetry in primary angle-closure glaucoma (PACG) as compared with that in primary open-angle glaucoma (POAG) and to compare the retinal nerve fiber layer (RNFL) distribution between PACG and POAG. Prospective, comparative, observational cases series. One eye each of 58 PACG patients and 51 POAG patients. Scanning laser polarimetry with variable corneal compensation (GDx VCC). GDx VCC temporal-superior-nasal-inferior-temporal (TSNIT) parameters, including TSNIT average, TSNIT standard deviation, superior average, and inferior average, as well as the nerve fiber indicator (NFI). By using a logistic marginal regression model that defined an abnormal test as P<5% for each of the TSNIT parameters or NFI > or = 31, we found that diagnostic sensitivities of the GDx VCC parameters were similar (all Ps>0.05) in PACG and POAG despite the differences in refraction error (P = 0.017), axial length (P<0.001), and disc diameters (vertical, P = 0.031; horizontal, P = 0.002) between these 2 forms of glaucoma. The between-group similarity in the diagnostic sensitivity remained true either when all eyes were considered together or in each severity group, based on the visual field scoring system adopted by the Advanced Glaucoma Intervention Study. Regarding the RNFL distribution, the parameter inferior average was greater than the superior average in either PACG (P = 0.010) or POAG (P = 0.006). Further subgroup analyses found significant superior-inferior asymmetry in mild PACG (P = 0.022) but not in mild POAG (P = 0.279). Eyes with PACG have ocular biometrics and, possibly, pathogeneses of optic nerve damage different from those of eyes with POAG; however, the diagnostic sensitivity of GDx VCC is quite comparable in these 2 forms of glaucoma, irrespective of disease severity.