Diagnostic Tools for Calculation of Glaucoma Risk - Ocular ...

 
CONTINUE READING
SURVEY OF OPHTHALMOLOGY             VOLUME 53  SUPPLEMENT 1  NOVEMBER 2008

Diagnostic Tools for Calculation of Glaucoma Risk
Steve L. Mansberger, MD, MPH,1 Felipe A. Medeiros, MD, PhD,2 and Mae Gordon, PhD3

1
 Devers Eye Institute, Legacy Health System, Portland, Oregon, USA; 2Shiley Eye Center, University of California, San
Diego, California, USA; and 3Department of Ophthalmology and Visual Sciences, Washington University, St. Louis,
Missouri, USA

          Abstract. Risk calculators may simplify the management of ocular hypertension and glaucoma
          patients and provide evidence-based treatment. Risk calculators are not new to medicine. Medical
          providers use risk calculators to guide decision-making for the risk of a chromosomal abnormality in
          neonates, osteoporosis in postmenopausal women, and cardiovascular disease in adults. This article
          describes the current knowledge of risk calculators in ophthalmology. Clinicians should recognize that
          the current risk calculators do not include critical information guiding treatment such as life
          expectancy and patient’s willingness to commit to years of eye drops. Overall, eye-care providers should
          consider the results of a risk calculator as supplemental information when treating an ocular
          hypertension or glaucoma patient. (Surv Ophthalmol 53:S11--S16, 2008. Ó 2008 Elsevier Inc. All
          rights reserved.)

          Key words. Early Manifest Glaucoma trial  European Glaucoma Prevention study 
          glaucoma  ocular      hypertension  Ocular Hypertension Treatment   Study  risk
          assessment  risk calculator

Risk calculators can be an innovative approach to                 a chromosomal abnormality when compared to the
simplify the management of ocular hypertension                    risk of amniocentesis. They use bone scans and
and glaucoma patients and provide evidence-based                  demographic factors to help decide whether post-
treatment. This aricle describes the what, where,                 menopausal women require treatment to prevent
how, and why of risk calculators in ophthalmology. It             osteoporosis. Another popular use occurs in internal
also identifies two Web sites to attain the latest risk           medicine. The Framingham Heart Study developed
calculators in several software versions.                         a predictive equation to estimate the risk of cardio-
                                                                  vascular disease with age, lipid levels, blood pressure,
                                                                  diabetes, and smoking history as explanatory vari-
          Risk Calculators in Medicine                            ables.14 Medical providers use these results to decide
   The Institute of Medicine’s book, Crossing the                 whether to treat high blood pressure and cholesterol.
Quality Chasm, recommends safe, effective, efficient,
and patient-centered health care in the 21st century.5
Eye-care providers need tools to practice this type of                 Using a Risk Calculator for Ocular
health care. A risk calculator may be such a tool.                               Hypertension
   Risk calculators are not new to medicine. Medical                In the United States, approximately 8% of adults
providers use demographic factors and prenatal blood              over the age of 40 years have ocular hypertension.13
tests to counsel pregnant women regarding the risk of             Whereas ocular hypertension is a common finding,

                                                            S11
Ó 2008 by Elsevier Inc.                                                                      0039-6257/08/$--see front matter
All rights reserved.                                                                    doi:10.1016/j.survophthal.2008.08.005
S12     Surv Ophthalmol 53 (Suppl 1) November 2008                                             MANSBERGER ET AL

eye-care providers do not always know which patients
to treat or which patients to monitor without
treatment. To address this issue, the Ocular Hyper-
tension Treatment Study (OHTS) and European
Glaucoma Prevention Study (EGPS) determined the
baseline factors that predict the onset of primary open-
angle glaucoma in ocular hypertensive patients.3
   Ocular hypertension for the combined OHTS and
EGPS analysis3 included patients who met the
following criteria: age between 30 to 80 years, IOP
between 20 mm Hg and 32 mm Hg in both eyes, best
corrected visual acuity of 20/40 or better, normal
                                                             Fig. 1. Scatterplot of ophthalmologists’ estimate of risk
automated visual fields, normal optic disks, and open        (%) and whether or not they would treat. The dashed line
angles on gonioscopy. Patients with secondary causes         within the figure represents the risk calculator estimate.
of high IOP (e.g., pseudoexfoliation, pigment disper-        Reprinted from Mansberger et al11 with permission of
sion syndrome, corticosteroid use, trauma), other            Journal of Glaucoma.
intraocular eye disease, history of refractive surgery, or
                                                             provides clinicians an individualized estimate of
other diseases possibly affecting the visual field (e.g.,
                                                             susceptibility to developing glaucoma in 5 years for
central nervous syndrome tumors) were excluded.
                                                             a patient; thus risk calculators encourage treatment
Patients with any evidence of diabetic retinopathy
                                                             to be patient-centered rather than population-
documented from a dilated ophthalmoscopic exam-
                                                             based. In other words, patients are more likely to
ination were also excluded from the study.
                                                             adhere to therapy if they have a more definitive
   Trying to decide whether to treat an ocular
                                                             expectation of risk, rather than something vague,
hypertension patient is complex without a risk
                                                             such as a higher or lower risk. Thus, risk calculation
calculator. The newest OHTS multivariate regres-
                                                             may strengthen the physician--patient relationship
sion contains five variables that are predictive of
                                                             and enhance compliance.
developing glaucoma from ocular hypertension:
                                                                Clearly, providers should avoid treating every
age, corneal thickness, IOP, pattern standard de-
                                                             ocular hypertension patient. A risk calculator can
viation (PSD), and vertical cup-to-disk (C/D) ratio. 3
                                                             help select those patients who will most benefit from
Even if one simplifies the continuous variables of
                                                             treatment because of their high risk of developing
age, corneal thickness, IOP, and PSD into thirds and
                                                             glaucoma. It can also determine which patients have
uses nine different combinations for C/D (0.0--0.8),
                                                             a low risk of developing glaucoma and should not be
729 (3  3  3  3  9) different results exist for
                                                             treated. Finally, when a calculator determines
ocular hypertension patients. This creates a large
                                                             a borderline risk, the provider’s experience and
number of different combinations, which is difficult
                                                             the patient’s input can provide guidance regarding
for clinicians to decipher when deciding whether to
                                                             whether to treat.
treat a particular ocular hypertension patient.
   This difficulty was highlighted in a recent publi-
cation that estimates an ophthalmologist’s ability to
predict the risk of glaucoma in ocular hypertensive            Benefits to Society of a Risk Calculator
patients.11 Ophthalmologists had the benefit of an              Risk calculators may simultaneously save money
oral review and written handouts summarizing the             and decrease blindness. For example, some pro-
OHTS results, but they tended to underestimate the           viders treat all ocular hypertensive patients based on
risk when compared to the actual risk found by               the OHTS study results. This produces excess costs
a risk calculator. They also had a large range of            to society. The 5-year cumulative probability of
predictions, sometimes differing from the actual risk        developing glaucoma was 9.5% in the untreated
by 40% (Fig. 1). In general, this study showed that          group and 4.4% in the treated group. This results in
eye-care providers may frequently over- or under-            a relative risk reduction of 54% ([9.5 -- 4.4] 4 9.5)
treat their ocular hypertensive patients because of          and an absolute risk reduction of 5.1% (9.5 -- 4.4).
the difficulty associated with risk assessment.              In this example, the number needed to treat (NNT)
                                                             is 20 (1/0.051) persons to prevent 1 person from
                                                             developing glaucoma. If one uses visual field loss
                                                             (an outcome more likely to be associated with
               Benefits to Patients                          decreased quality of life) as the criteria for
  A risk calculator may provide benefits to patients         glaucoma, the NNT increases to 42 persons. The
regarding compliance. The OHTS risk calculator               main reason for these high NNT values is the low
DIAGNOSTIC TOOLS FOR CALCULATION OF GLAUCOMA RISK                                                           S13

risk of developing glaucoma for the majority of          hypertension who were followed as part of a pro-
ocular hypertension patients.                            spective longitudinal study conducted at the Uni-
   The excess 5-year cost to prevent one ocular          versity of California San Diego (DIGS; Diagnostic
hypertensive patient from developing glaucoma is         Innovations in Glaucoma Study).
US$ 54,000 with an NNT of 20 and US$ 113,000 with           Several steps were taken to validate the OHTS-
an NNT of 42 (assuming effective monotherapy with        derived model. In the first step, the importance of
a $20 beta-adrenergic antagonist and two extra           the prognostic variables that had been previously
office visits per year). Substituting an $80 prosta-     identified by the OHTS study was evaluated on the
glandin analog for treatment increases the cost to       new data set (DIGS data set). All the variables had
$126,000 with an NNT of 20 and $264,600 with an          similar performance, except for diabetes mellitus,
NNT of 42. Clinicians need to consider these costs as    which was not significantly associated with the risk of
well as side effects, efficacy, and convenience when     developing glaucoma in the DIGS data. Subse-
choosing an ocular hypotensive medication for            quently, the predictive performance of the model
treatment.                                               was investigated on the new data set. The ability of
   An economic analysis6 has shown that not all          the OHTS-derived risk calculator to discriminate
patients require treatment of ocular hypertension,       DIGS subjects who developed glaucoma from those
just those with higher risk; this results reported       who did not was reasonably good with a c-index of
showed a probability of developing glaucoma of           approximately 0.7. The c-index is a measure of the
greater than 2% per year (or 10% over 5 years) as        discriminating ability of a model (similar to the area
the threshold for cost-effective treatment of ocular     under the receiver operating characteristic [ROC]
hypertension patients.                                   curve) and a c-index of 0.7 indicates that, in
                                                         approximately 70% of the cases, the model allocated
                                                         a higher predicted probability for a subject who
                                                         actually developed glaucoma than for a subject who
        Validation of Risk Calculators                   did not. The closer the c-index gets to 1, the better
   The development of risk calculators involves use      the discriminating ability of the model. The values
of statistical methods to develop models for pre-        of c-index found for the OHTS-derived risk calcu-
diction of outcome using one or more explanatory         lator when applied to DIGS subjects were similar to
variables. However, a model that is derived from         those found when risk models such as the Framing-
a particular data set is not guaranteed to work on       ham coronary prediction scores are used to predict
a different group of patients. In fact, the perfor-      coronary heart disease events.3,10 D’Agostino et al
mance of regression models (or risk calculators)         reported c-indexes ranging from 0.63 to 0.83 when
used as diagnostic or prediction tools is generally      the Framingham functions were applied to six
better on the data set on which the model has been       different cohorts of patients.3
constructed (derivation set) compared to the                The OHTS-derived risk calculator also had a good
performance of the same model on new data.               calibration when applied to the DIGS data set
Therefore, before risk calculators can be successfully   (Fig. 2). Checking calibration is another important
incorporated into clinical practice they need to be      step in validating a predictive model. A reliable or
validated on different populations. By validation we     well-calibrated model will give predicted probabili-
mean establishing that the risk calculator works         ties that agree numerically with the actual outcomes.
satisfactorily for patients other than those from        For example, suppose a group of 100 ocular
whose data the model was derived.                        hypertensive patients. If the model assigns an
   In 2005, we published the results on the de-          average probability of 12% for conversion to
velopment and validation of a risk calculator to         glaucoma for this group of subjects, it is expected
assess the risk of an ocular hypertensive patient to     that approximately 12 subjects will convert to
develop glaucoma.12 The risk calculator was derived      glaucoma over time. That is, for a well-calibrated
based on the results published by the OHTS,2,3 and       model, the predicted probabilities of conversion to
it incorporated the variables that were described in     glaucoma will agree closely with the observed
that study as being significantly associated with the    probabilities of conversion. Fig. 2 shows that the
risk of developing glaucoma over time—that is, age,      OHTS-derived risk calculator performed well on the
IOP, CCT, PSD, vertical C/D ratio, and diabetes          DIGS data set. For patients in whom the model
mellitus. The risk calculator was designed to            predicted a high chance of converting to glaucoma,
estimate the chance that an ocular hypertensive          there was a high observed conversion rate; whereas
patient would develop glaucoma if left untreated for     for patients in whom the model predicted a low
5 years. It was subsequently validated on an in-         conversion rate, there was a low observed conversion
dependent population of 126 patients with ocular         rate.
S14                                          Surv Ophthalmol 53 (Suppl 1) November 2008                                           MANSBERGER ET AL

                                                                                                obtain the risk calculator free of charge at http://
                                      0.40                                          Observed
                                                                                    Predicted
                                                                                                ohts.wustl.edu/risk (Web-based and Adobe Acrobat)
                                                                                                and www.deverseye.org/grc_web/grc.cfm (Web-based,
Probability of Glaucoma Development

                                                                                                Palm, Cþþ, Pocket PC and MacIntosh versions). As
                                                                                                new factors for the development of glaucoma are
                                      0.30
                                                                                                identified, the risk calculator will need to be modified.

                                      0.20
                                                                                                       Risk Assessment for Progressive
                                                                                                                  Glaucoma
                                                                                                   The Early Manifest Glaucoma Trial (EMGT)
                                                                                                randomized patients with early glaucoma either to
                                      0.10                                                      argon laser trabeculoplasty plus betaxolol (n 5 129)
                                                                                                or to monitoring without immediate treatment (n 5
                                                                                                126).8 These were newly diagnosed glaucoma
                                                                                                patients, found during a community glaucoma
                                      0.00                                                      screening program. The rate of progression was
                                                  1          2           3             4        45% in the treated group versus 62% in the
                                                       Quartile of Predicted Risk               untreated group. The treatment reduced IOP by
Fig. 2. Comparison of observed and predicted 5-year
                                                                                                approximately 20% and decreased the risk of
incidence of glaucoma when the OHTS-derived risk                                                worsening glaucoma by 50%.7 The study identified
calculator was applied to the DIGS (Diagnostic Innova-                                          elevated IOP, exfoliation, bilateral disease, worse
tions in Glaucoma Study) data set. Predicted probabilities                                      mean deviation with perimetry, and older age as
agreed closely with observed outcomes. Reprinted from                                           baseline risk factors for glaucoma progression, and
Medeiros et al 12 with permission of Archives of
Ophthalmology.
                                                                                                central corneal thickness (CCT) was not found to be
                                                                                                a risk factor. During follow-up, disk hemorrhages
                                                                                                increased the risk of progression.
   In 2007, we validated the OHTS derived risk
                                                                                                   The EMGTrecently published the predictors of long-
calculator in a large independent European sample
                                                                                                term progression (mean 8 years) in patients with early
of patients with ocular hypertension.3 These analyses
                                                                                                glaucoma.9 Their results are similar to those reported
were based on nearly 5 years of follow-up data from
                                                                                                previously, but now showed an increased risk of
717 OHTS patients who did not receive hypotensive
                                                                                                glaucomatous progression with thin CCT and de-
treatment and 406 patients in the placebo group of
                                                                                                creased perfusion pressure. The results were interest-
the European Glaucoma Prevention Study (EGPS).
                                                                                                ing in that statistical interaction between CCT and IOP
The European sample provided the strongest
                                                                                                occurred with CCT only a risk factor in those with an
evidence to date of the generalizeability of the
                                                                                                IOP O21, not those with IOP !21. Overall, interpre-
OHTS-derived risk calculator to other patient
                                                                                                tation of these results is more difficult than with OHTS,
populations with ocular hypertension. Not only did
                                                                                                and the study has a smaller sample size. Therefore,
independent analyses of the EGPS data identify the
                                                                                                a risk assessment equation may be less precise.
same predictors that increased the risk of POAG
                                                                                                   An EMGTrisk calculator could identify the patients
(older age, higher IOP, thiner CCT, larger C/D
                                                                                                at highest risk for glaucomatous progression. These
ratio, and higher PSD) , but the risk coefficients of
                                                                                                patients could be monitored closely and treated more
each of these predictors did not differ from each
                                                                                                aggressively when compared to a patient who is
other (p values of 0.53, 0.49, 0.89, 0.96, and 0.55,
                                                                                                unlikely to progress. Finally, risk calculators could
respectively). In these pooled analyses, diabetes was
                                                                                                identify those who would benefit from a specific type
not found to be a risk factor for the development of
                                                                                                of treatment, such as ocular hypotensive medications,
POAG and was dropped from the risk model. The
                                                                                                argon laser trabeculoplasty, and surgery. Overall, this
risk model from the pooled OHTS/EGPS sample of
                                                                                                may decrease the risk of blindness in glaucoma
over 1,100 ocular hypertensive patients demon-
                                                                                                patients. The authors of the EMGT have not yet
strated excellent fit with the observed data with
                                                                                                released a risk assessment equation.
a c-statistic of 0.74 and the calibration statistic was
7.05.3 The large sample substantially increases the
statistical precision of coefficient estimates and
thereby reduces the errors of estimating risk.                                                            Caveats of Risk Calculators
   To simplify application of the risk model to an                                                A recent study suggested that in a critical care
individual patient with ocular hypertension, one can                                            setting, clinicians may not change their treatment
DIAGNOSTIC TOOLS FOR CALCULATION OF GLAUCOMA RISK                                                                    S15

based on a risk calculator.1 This randomized, clinical     treatment, such as medical health and life expec-
trial showed that even when a risk model predicted         tancy, patient’s willingness to commit to years of eye
an intensive care unit patient would die within            drops, cost, and the effect of quality of life with
a week, doctors rarely used this information to            treatment. Overall, eye care providers should con-
obtain an end-of-life recommendation from the              sider the results of a risk calculator as supplemental
family.1 The authors suggested that the doctors were       information when treating ovular hypertension or
unwilling to apply results from the calculator due to      glaucoma patient.
‘‘inertia’’ and ‘‘lack of incentives’’ with the current
situation. We believe that implementation of risk
assessment would differ between a critical care
                                                                                 Conclusion
setting and an ophthalmologist’s office, but pro-             Predicting the development of glaucoma from
viding a risk assessment tool does not guarantee           ocular hypertension is a cornerstone to deciding on
adoption by clinicians.                                    whether or not to treat. Risk calculators may be an
   Risk assessment and calculation has several other       innovative approach to simplify the management of
limitations. Risk calculators are based on the best        ocular hypertension and glaucoma patients and
available information, thus their use should be            provide evidence-based treatment. They provide
restricted to patients that are similar to the inclusion   benefits to patients, clinicians, and society as a whole.
criteria of the study. For example, with regards to        However, eye care providers should recognize that
a risk calculator based on the OHTS, ophthalmol-           risk assessment is still evolving and requires re-
ogists should not assume that eyes with secondary          finement. Therefore, eye care providers should
causes of ocular hypertension, such as pseudoexfo-         consider the result of a risk calculator as supplemen-
liation, juvenile open-angle glaucoma, chronic             tal information when deciding whether to treat.
angle-closure glaucoma, pigmentary dispersion syn-
drome, and so on, will have similar risk factors to the
OHTS population. A risk calculator can be impre-                   Methods of Literature Search
cise if the representative patient has a rare combi-          The authors performed a PubMed search using
nation of characteristics such as diabetes, a smaller      the terms ocular hypertension, glaucoma, risk calculator,
C/D ratio, a thicker cornea, and older age. This rare      risk assessment, prediction model, from the period 1970
combination creates a lack of similar patients,            until 2007. They included pertinent articles from
inadequate sample size for comparison, and a large         their personal collection if not contained within the
confidence interval for the log odds.                      above search. Their search did not include any
   Clinicians need to understand that a risk calcula-      articles written in a language other than English.
tor provides a mean risk based on a group of
patients with similar characteristics. Rare combina-
tions in OHTS, such as a C/D ratio less than 0.2, an                             References
IOP above 29 mm Hg, or an age of above 70 years,
                                                           1. ___: A controlled trial to improve care for seriously ill
will result in larger confidence intervals around             hospitalized patients. The study to understand prognoses
individual estimates and therefore less precise               and preferences for outcomes and risks of treatments
estimates.4                                                   (SUPPORT). The SUPPORT Principal Investigators. JAMA
                                                              274:1591--8, 1995
   The patients who were enrolled in OHTS may not          2. Coleman AL, Gordon MO, Beiser JA, et al: Baseline risk
accurately represent a typical ocular hypertension            factors for the development of primary open-angle glau-
patient in an eye care provider’s office, who may             coma in the Ocular Hypertension Treatment Study. Am J
                                                              Ophthalmol 138:684--5, 2004
have a lower or higher probability of developing           3. Gordon MO, Torri V, Miglior S, et al: Validated prediction
glaucoma. These differences may occur because the             model for the development of primary open-angle glaucoma
OHTS patients received free medications and                   in individuals with ocular hypertension. Ophthalmology
                                                              114:10--9, 2007
clinical visits, were reliable visual field takers, and    4. Hosmer DW, Lemeshow S: Applied survival analysis: re-
were apparently compliant with their medications              gression modeling of time to event data, Vol 13. New York,
and follow-up. These characteristics are typical of           Wiley, 1999. p 386
                                                           5. Institute of Medicine (U.S.): Committee on Quality of
research study patients, but uncommon in the                  Health Care in America: Crossing the quality chasm: a new
clinical realm. Thus, validation in additional patient        health system for the 21st century. Washington, DC,
populations and everyday clinical settings would              National Academy Press, 2001. p 337
                                                           6. Kymes SM, Kass MA, Anderson DR, et al: Management of
enhance the ability to generalize the risk calculators.       ocular hypertension: a cost-effectiveness approach from the
   Finally, clinicians should use more than the               Ocular Hypertension Treatment Study. Am J Ophthalmol
results of a risk calculator to decide whether or             141:997--1008, 2006
                                                           7. Leske MC, Heijl A, Hussein M, et al: Factors for glaucoma
not to treat a patient. The current risk calculators          progression and the effect of treatment: the early manifest
do not include important information guiding                  glaucoma trial. Arch Ophthalmol 121:48--56, 2003
S16      Surv Ophthalmol 53 (Suppl 1) November 2008                                                        MANSBERGER ET AL

 8. Leske MC, Heijl A, Hyman L, et al: Early Manifest Glaucoma      13. Tielsch JM, Katz J, Singh K, et al: A population-based
    Trial: design and baseline data. Ophthalmology 106:2144--           evaluation of glaucoma screening: the Baltimore Eye Survey.
    53, 1999                                                            Am J Epidemiol 134:1102--10, 1991
 9. Leske MC, Heijl A, Hyman L, et al: Predictors of long-term      14. Wilson PW, D’Agostino RB, Levy D, et al: Prediction of
    progression in the early manifest glaucoma trial. Ophthal-          coronary heart disease using risk factor categories. Circula-
    mology 114:1965--72, 2007                                           tion 97:1837--47, 1998
10. Liu J, Hong Y, D’Agostino RB, et al: Predictive value for the
    Chinese population of the Framingham CHD risk assess-
    ment tool compared with the Chinese Multi-Provincial              The supplement in which this article is published was funded
    Cohort Study. JAMA 291:2591--9, 2004                            by Pfizer. The authors reported no proprietary or commercial
11. Mansberger SL, Cioffi GA: The probability of glaucoma           interest in any product mentioned or concept discussed in this
    from ocular hypertension determined by ophthalmologists         article. This article was supported in part by grants from NEI 3
    in comparison to a risk calculator. J Glaucoma 15:426--31,      K23 EY0155501-01 (S.L.M.), NEI EY09341, EY09307 (M.G.),
    2006                                                            Merck Research Laboratories (M.G.), and Research to Prevent
12. Medeiros FA, Weinreb RN, Sample PA, et al: Validation of        Blindness (M.G.).
    a predictive model to estimate the risk of conversion from        Reprint address: Steven L. Mansberger, MD, MPH. Devers Eye
    ocular hypertension to glaucoma. Arch Ophthalmol 123:           Institute/Discoveries in Sight, 1040 NW 22nd Avenue, Suite 200,
    1351--60, 2005                                                  Portland, OR 97210. e-mail: smansberger@deverseye.org.
You can also read