Dataset Information

Filtering data from the collaborative initial glaucoma treatment study for improved identification of glaucoma progression.

ABSTRACT:

Background

Open-angle glaucoma (OAG) is a prevalent, degenerate ocular disease which can lead to blindness without proper clinical management. The tests used to assess disease progression are susceptible to process and measurement noise. The aim of this study was to develop a methodology which accounts for the inherent noise in the data and improve significant disease progression identification.

Methods

Longitudinal observations from the Collaborative Initial Glaucoma Treatment Study (CIGTS) were used to parameterize and validate a Kalman filter model and logistic regression function. The Kalman filter estimates the true value of biomarkers associated with OAG and forecasts future values of these variables. We develop two logistic regression models via generalized estimating equations (GEE) for calculating the probability of experiencing significant OAG progression: one model based on the raw measurements from CIGTS and another model based on the Kalman filter estimates of the CIGTS data. Receiver operating characteristic (ROC) curves and associated area under the ROC curve (AUC) estimates are calculated using cross-fold validation.

Results

The logistic regression model developed using Kalman filter estimates as data input achieves higher sensitivity and specificity than the model developed using raw measurements. The mean AUC for the Kalman filter-based model is 0.961 while the mean AUC for the raw measurements model is 0.889. Hence, using the probability function generated via Kalman filter estimates and GEE for logistic regression, we are able to more accurately classify patients and instances as experiencing significant OAG progression.

Conclusion

A Kalman filter approach for estimating the true value of OAG biomarkers resulted in data input which improved the accuracy of a logistic regression classification model compared to a model using raw measurements as input. This methodology accounts for process and measurement noise to enable improved discrimination between progression and nonprogression in chronic diseases.

SUBMITTER: Schell GJ

PROVIDER: S-EPMC3878032 | biostudies-literature | 2013 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:PurposeTo evaluate the relationship between medication adherence and visual field progression in participants randomized to the medication arm of the Collaborative Initial Glaucoma Treatment Study (CIGTS).DesignThe CIGTS was a randomized, multicenter clinical trial comparing initial treatment with topical medications to trabeculectomy for 607 participants with newly diagnosed glaucoma.ParticipantsThree hundred seven participants randomized to the medication arm of the CIGTS.MethodsParticipants were followed up at 6-month intervals for up to 10 years. Self-reported medication adherence and visual fields were measured. Medication adherence was assessed by telephone from responses to the question, "Did you happen to miss any dose of your medication yesterday?" The impact of medication adherence on mean deviation (MD) over time was assessed with a linear mixed regression model adjusting for the effects of baseline MD and age, cataract extraction, interactions, and time (through year 8, excluding time after crossover to surgery). Medication adherence was modeled as a cumulative sum of the number of prior visits where a missed dose of medication was reported.Main outcome measureMean deviation over time.ResultsThree hundred seven subjects (306 with adherence data) were randomized to treatment with topical medications and followed up for an average of 7.3 years (standard deviation, 2.3 years). One hundred forty-two subjects (46%) reported never missing a dose of medication over all available follow-up, 112 patients (37%) reported missing medication at up to one third of visits, 31 patients (10%) reported missing medication at one third to two thirds of visits, and 21 patients (7%) reported missing medication at more than two thirds of visits. Worse medication adherence was associated with loss of MD over time (P = 0.005). For subjects who reported never missing a dose of medication, the average predicted MD loss over 8 years was 0.62 dB, consistent with age-related loss (95% confidence interval [CI], 0.17-1.06; P = 0.007); subjects who reported missing medication doses at one third of visits had a loss of 1.42 dB (95% CI, 0.86-1.98; P < 0.0001); and subjects who reported missing medication doses at two thirds of visits showed a loss of 2.23 dB (95% CI, 1.19-3.26; P < 0.0001).ConclusionsThis longitudinal assessment demonstrated a statistically and clinically significant association between medication nonadherence and glaucomatous vision loss.

Project description:ObjectiveTo evaluate the impact of measures of intraocular pressure (IOP) control on progression of visual field (VF) loss during long-term treatment for open-angle glaucoma (OAG).DesignLongitudinal, randomized clinical trial.ParticipantsWe included 607 participants with newly diagnosed OAG.MethodsStudy participants were randomly assigned to initial treatment with medications or trabeculectomy, and underwent examination at 6-month intervals. Standardized testing included Goldmann applanation tonometry and Humphrey 24-2 full threshold VFs. Summary measures of IOP control during follow-up included the maximum, mean, standard deviation (SD), range, proportion less than 16, 18, 20, or 22 mmHg, and whether all IOP values were less than each of these 4 cutpoints. Predictive models for VF outcomes were based on the mean deviation (MD) from VF testing, and were adjusted for age, gender, race, baseline VF loss, treatment, and time. Each summary IOP measure was included as a cumulative, time-dependent variable, and its association with subsequent VF loss was assessed from 3 to 9 years postrandomization. Both linear mixed models, to detect shifts in MD levels, and logistic models, to detect elevated odds of substantial worsening (?3 dB), were used.Main outcome measuresWe measured the MD from Humphrey 24-2 full threshold VF tests.ResultsThe effect of the summary IOP measures differed between the medicine and surgery groups in models that addressed the continuous MD outcome. After adjustment for baseline risk factors, in the medicine group larger values of 3 IOP control measures-maximum IOP (P = 0.0003), SD of IOP (P = 0.0056), and range of IOP (P<0.0001)-were significantly associated with lower (worse) MD over the 3- to 9-year period. No IOP summary measure was significantly associated with MD over time in the surgery group. The same 3 IOP summary measures were also significantly associated with substantial worsening of MD; however, the effects were similar in both treatment groups. In models predicting inadequate IOP control, consistently significant predictors of higher maximum, SD, and range of IOP included black race, higher baseline IOP, and clinical center.ConclusionsThese results support considering more aggressive treatment when undue elevation or variation in IOP measures is observed.

Project description:ImportanceClinicians would benefit from knowing the association of glaucomatous change in 2 eyes of the same patient.ObjectiveTo estimate the time between initial glaucoma treatment of the study eye (SE) and the need for treatment of the fellow eye (FE) and to ascertain the concordance between rates of progression in SEs and FEs.Design, setting, and participantsIn this post hoc analysis of participants from the Collaborative Initial Glaucoma Treatment Study, 607 participants with newly diagnosed open-angle glaucoma in 1 or both eyes were followed up for up to 11 years. An SE was designated at baseline and randomized to medical or surgical treatment. By protocol, FEs were treated when eligible or at physician discretion. Survival analysis methods were used to estimate the probability of FE treatment over time and to test baseline and time-dependent predictors of treatment. Disease trajectory was calculated with linear regression as the patient eye-specific slopes of mean deviation (MD) and intraocular pressure (IOP) over time, and correlations between SE and FE trajectories were calculated. Data were collected from October 1993 to December 2004. Data were analyzed from September 2012 to May 2018.Main outcomes and measuresTime to FE treatment and slopes over time of MD and IOP in SEs and FEs.ResultsOf the 607 included patients, 334 (55.0%) were male and 337 (55.5%) were white, and the mean (SD) age was 58.0 (10.9) years. A total of 291 FEs (47.9%) were treated at baseline, 123 (20.3%) were eventually treated, and 193 (31.8%) never received treatment. The probability of FE treatment for open-angle glaucoma was 0.57 at 1 year and 0.68 at 7 years after randomization. Characteristics significantly associated with an increased hazard of FE treatment included older age (hazard ratio [HR], 1.33; 95% CI, 1.08-1.64; P = .007), hypertension (HR, 1.76; 95% CI, 1.16-2.67; P = .008), higher IOP (HR, 1.24; 95% CI, 1.20-1.29; P < .001), large cup-disc ratio (HR, 1.40; 95% CI, 1.23-1.58; P < .001), and worse MD. Correlations in MD slopes between SEs and FEs initially, eventually, and never treated were 0.73, 0.71, and 0.34, respectively. Comparable correlations in IOP slopes were 0.57, 0.24, and 0, respectively.Conclusions and relevanceAmong patients with newly diagnosed open-angle glaucoma, almost half had initial treatment of both eyes. After 7 years, approximately two-thirds of patients had bilateral treatment. Of the variables predictive of FE treatment, modifiable factors included hypertension and IOP. Slopes of MD were similar between SEs and treated FEs. This implies that SE change is a harbinger of FE change and therefore warrants close surveillance.

Project description:ImportanceAdministration of a brief questionnaire to evaluate glaucoma symptoms would lend useful information for patient care.ObjectivesTo develop a shortened glaucoma symptom measure based on the Collaborative Initial Glaucoma Treatment Study (CIGTS) Symptom and Health Problem Checklist (SHPC) and evaluate its psychometric properties.Design, setting, and participantsThis measure development study evaluated the factor structure of the 43-item SHPC that was obtained from CIGTS participants at baseline and every 6 months thereafter. These 607 participants were enrolled at 14 clinical centers in the United States and had newly diagnosed open-angle glaucoma. Their mean deviation (SD) from visual field testing was ?5.5 (4.3) dB. Data were collected from October 1993 through April 1997.Main outcomes and measuresThe factor structure of the SHPC, confirmatory factor analysis of the resulting 18-item questionnaire (SHPC-18), the reliability of the SHPC-18, and associations of the 2 symptom subscales (Local Eye and Visual Function) of the SHPC-18 with visual field severity and 25-item National Eye Institute Visual Function Questionnaire (NEI-VFQ25) subscales.ResultsAmong the 607 participants (mean [SD] age at enrollment, 57.5 [10.9] years), 334 (55.0%) were men and 273 (45.0%) were women; 231 (38.1%) were African American. Exploratory factor analysis and longitudinal growth modeling documented Local Eye and Visual Function symptom subscales. Cronbach ? values for mean weighted internal consistency were 0.83 and 0.89 for the Local Eye and Visual Function subscales, respectively, and remained stable over time. Scores on each subscale significantly correlated with the NEI-VFQ25 total score (r?=??0.41 and r?=??0.59, respectively) and with all subscale scores (P?<?.01). Participants with more severe glaucoma had higher (worse) mean (SD) scores than those with mild glaucoma at baseline on the Local Eye (4.68 [6.62] vs 3.07 [5.60]) and Visual Function (8.44 [11.45] vs 4.42 [8.94]; P?<?.05) SHPC-18 subscales. Participants who underwent trabeculectomy reported a higher frequency of any Local Eye symptoms than those treated with topical medications (eg, at 12 months, 153 of 269 [56.9%] vs 11 of 276 [40.9%]; P?<?.001).Conclusions and relevanceThese results suggest that the SHPC-18 is a reliable, responsive, and psychometrically sound measure of patient-reported, glaucoma-related symptoms. The measure is responsive to treatment and discriminates the severity of glaucoma. This shorter version of the original SHPC measure may be useful in clinical and research settings to better understand the influence of glaucoma symptoms on patients’ daily life.