The Berkeley Contact Lens Extended Wear Study: part I

Reprinted fbm Volume 108, Number 8, Auguss 2001

A M E R I C A N A C A D E M Y b

O F OPHTHALMOLOGY ~ ~ M . R A s ~

Ophthalmology The Berkeley Contact Lens Extended Wear Study: Part I Study Design and Conduct

Robert E. Fusaro, PhD, ID,' Kenneth A. Polse, OD, M S ~ , Andrew D. Graham, M S , ~ Cheslyn M. Can, OD,* Reuben K. Rivera, OD, MS, Meng C. Lin, OD, Timothy L. Sanders, OD ,2

Nancy A. McNarnara, OD, PhD,2 ]Judy S. Chan, OD, M S ~ Objective: The primary aim of the Berkeley Contact Lens Extended Wear Study (CLEWS) was to test the

hypotheses that extended wear of rigid gas-permeable (RGP) contact lenses with greater oxygen permeability (Dk) reduces the incidence of contact lens-associated keratopathy (CLAK) and increases the survival rate in RGP extended wear (EW). In this article we describe the clinical trial design in detail, present the results of subject recruitment and retention, and provide the baseline demographic and ocular characteristics of the CLEWS subjects, whose data will be analyzed to address the study aims in a companion article.

u Design: A randomized, concurrently controlled clinical trial. Intervention: Subjects were fitted with day wear (DW) high-Dk RGP lenses and then adapted to EW.

Subjects who adapted to EW were then randomly assigned to either high- or medium-Dk RGP lenses for 12 months of 6-nightdweek EW.

Main Outcome Measures: Slit-lamp assessment and grading of 17 possible keratopathies, measurement of refractive error and corneal curvature, and symptoms. Follow-up data were collected every 3 months.

Results: From 545 subjects entering the DW adaptation phase, 201 adapted to EW and were randomly assigned to medium- or high-Dk lenses for 12 months of EW. The baseline characteristics of the two study groups were similar and did not differ from the 344 DW subjects who failed to adapt to EW. The distributions of oxygen transmissibility for the two study groups were disjoint, indicating that each group received distinctly different levels of hypoxia.

Conclusions: We show that CLEWS was appropriately designed to address the study hypotheses, was conducted with regard for the safety of the subjects, and adhered to rigorous protocols designed to control for bias and ensure the integrity of study data. We establish the internal validity of between-group statistical comparisons and characterize our study population to permit informed evaluation of the applicability of our results to the contact lens-wearing population in general. Ophthalmology 2001;108:1381-1388 O 2001 by the American Academy of Ophthalmology.

Originally received: August 25, 2000. Accepted: March 9. 2001. Manuscript no. 200498.

' Division of Public Health Biology and Epidemiology, School of Public Health, University of California. Berkeley. California. 'Morton D. Sarver Laboratory for Cornea and Contact Lens Research. School of Optometry. University of California. Berkeley, California.

Supported by the National Eye Institute, National Institutes of Health, Bethesda. Maryland, EY07728: Concise Contact Lens Co.. San Leandro.

L California; and Paragon Vision Sciences. Mesa, Arizona.

Reprint requests to Kenneth A. Polse, OD, MS. School of Optometry. University of California. Berkeley, CA 94720-2020.

Contact lens (CL) extended wear has been associated with the development of various corneal complications, some of which can cause permanent visual impairment. I" Concerns about the risks assocjated with CL wear have resulted in considerable research aimed at ameliorating CL-associated keratopathy. Several studies have shown an association between wearing lenses known to cause corneal hypoxia and the occurrence of adverse corneal events such as increased epithelial cell loss,' inflammation,' microbial keratitis,'.* acute red eye,4 and endothelial polyrnegethism."

8 2001 by the American Academy of Ophthalmology Published by Elsevier Science Inc.

ISSN 016 1 -6420/01R-see front matter 1381 PI1 SO 16 1 -6420(0 1 W643- I

Ophthalmology Volume 108, Number 8, August 2001

Under controlled laboratory testing conditions, decreased lens oxygen transmissibility (Dklt) has been associated with increased corneal swel~ing, ' .~ increased binding of Pseudo- moms ueruginosu to epithelial cells (Cavanagh HD. Ren H, Ladage PM, et al. Invest Ophthalmol Vis Sci 405906 , 1999), decreased desquamation rates of epithelial cells? impaired corneal volume regulation, and lowered corneal pH. ' '

On the basis of these findings, many clinicians and researchers have assumed that the incidence and severity of CL-related adverse events depend in part o n the hypoxic dosage (i.e., oxygen tension at the tear-lens interface) that the cornea receives during overnight lens wear. Although these studies provide evidence for a hypoxic mechanism, further research is required to clearly demonstrate a link between hypoxic dosage and either the incidence of complications o r the ability to maintain successful C L extended wear. Determining whether there is a dose-response relationship between the risk of keratopathy and corneal hypoxia, or whether, for example, the risk is uniformly increased given some threshold hypoxic level, has implications for understanding the mechanisms associated with CL-associated keratopathy, as well as for the design and clinical use of CLs.

This uncertainty about the exact relationship between hypoxia and adverse ocular response in overnight C L wear prompted our corneal research group to undertake the Berkeley Contact Lens Extended Wear Study (CLEWS). CLEWS was a randomized, concurrently controlled clinical trial involving comprehensive quarterly follow-up of more than 200 subjects randomly assigned to wear either medium- or high-oxygen permeable (Dk) rigid gas-permeable (RGP) lenses for 12 months of extended wear (EW). The primary aims of CLEWS were to determine whether wearing high oxygen-permeable RGP lenses for extended wear (1) lowers the incidence of CLassociated keratopathy, and

Table 1. Contact Lens Extended Wear Study Eligibility Criteria at Date of lnitial Orientation. (Subjects Who Met These Screening Criteria Were Then Scheduled for a Clinical

Examination and Diagnostic Lens Fitting to Determine Final . Eligibility to Participate in Contact Lens Extended Wear Study

and to Enter the Lens Wear Adaptation Phase [Stage 21).

1. Age: 18 to 39 2. Ocular health: normal 3. Refractive error: +7.00 to -9.00 diopter spherical, and 0.00 to - 1.75 dioptcr cylindric

4. Residence: San Francisco Bay Area for at least rhe next 1.5 yrs 5. Ocular or mcdical history: no conditions or injuries thar might

adversely affect ocular health or otherwise limit ability to wear contact Lenses

Corneal surgery, corneal trauma Medications, including antidepressants, steroids Allergies with pronounced ocular effects (e.g., tearing, itching)

6. History of overnight polymethyl mcthacrylate contact lens wear: none

7. History of or current daily contact lens wear: minimal or none < 6 months total wearing time, or Currently wearing, or discontinued 5 6 mos ago

Worn C: 12 mos, < 30 hslwk average Worn > 12 mos, < 15 hslwk average

Discontinued wear > 6 mos ago Worn 5 24 mos, < 30 hslwk average Worn > 24 mos, < 15 hslwk avenge

Material and Methods

CLEWS was designed with three main stages: screening and baseline evaluation (stage I), adaptation to CL wear (stage 2), and randomized lens wear and follow-up (stage 3). We first describe . these three stages in detail, outline the bias control measures used in CLEWS, and describe the changes in protocol made during the course of the trial. We then describe the study lenses, personnel, and procedures, including data management and quality assurance and external interim review of CLEWS conduct by a Data and Safety Monitoring Board (DSMB).

- - (2) reduces the rate at which subjects discontinue overnight The CLEWS Clinical Trial Design wear because of lens-related com~licat ions. Secondarv aims of the study were to assess the effects of hypoxic dosage on various laboratory measures of corneal structure and function, to identify possible predictors of R G P EW-associated keratopathy, and to develop a semiautomated data management and quality assurance system for clinical research. CLEWS lasted for approximately 5 years, from the beginning of recruitment in April 1994 to the end of data collection in July 1999.

The purposes of this first article are to ( I ) present the main features of the CLEWS clinical trial design; (2) describe the recruitment, measurement, and bias control strategies; and (3) present the results of subject recruitment, retention, and baseline assessments, which are necessary to define the CLEWS study population and permit interpreta- tion of subsequent statistical results. W e analyze in a companion article (Polse KA, Graham AD, Fusaro RE, et al, The Berkeley Contact Lens Extended Wear Study. Part 11: Clinical Results) the clinical outcomes to address the primary study aims. Secondary aims of CLEWS will be ad- dressed in several other separate articles.

Stage 1. Screening and Baseline Evaluation

Prospective subjects were recruited from the San Francisco Bay Area and the general University of California at Berkeley (UCB) community through newspaper advertisements, fliers, UCB School of Optometry clinic referrals, and direct referrals. All subjects attended an orientation meeting at which information about age, gender. ethnicity, occupation, medical history, and previous CL experience was collected, the current spectacle prescription measured, and keratometry readings obtained. Ta- ble 1 lists the eligibility criteria used in stage I to accept or reject prospective subjects.

Because some types of CL wear have been shown to affect corneal structure and f ~ n c t i o n , ' ~ . ' ~ we initially attempted to enroll only subjects with no previous CL experience; however, unantic- ipated difficulties in enrolling eligible subjects for full participation in CLEWS prompted us to accept subjects with some minimal CL experience (as defined in Table 1). No prior polymethyl- methacrylate or overnight CL wear was permitted. Day wear of RGP or soft lenses was permitted if the subject had worn CLs for less than 6 months total wearing time. For subjects with longer total time in CL wear, a set of rules was applied that accounted for - how recently and how many hours per week, on average, lenses

Fusaro et a1 . Berkeley Contact Lens Extended Wear Study: Design

were worn. We believe these criteria reflected minimal CL experience that was unlikely to have had a negative impact on the cornea before participation in CLEWS. - Subjects who met the initial screening criteria were then given a detailed explanation of the study procedures, followed by a comprehensive eye and CL fitting examination. Those subjects meeting all study eligibility criteria then proceeded to stage 2. Informed consent, as approved by the UCB Committee for Pro- tection of Human Subjects, was obtained from each participant.

(e.g., to monitor development of subclinical keratopathy observed at a regular quarterly visit). During any phase of the study, subjects could elect to discontinue participation or be dropped from the study by decision of the clinician and principal investigator (PI). At the conclusion of the study, subjects were asked to discontinue lens wear for 1 week and report for a final clinical assessment.

Bias Control Measures

Stage 2. Adaptation to Contact Lens Wear

In stage 2, subjects adapted first to day wear of high-Dk RGP lenses and, if successful, then proceeded to EW adaptation, fol- lowing standard clinical procedures.13 All subjects wore the high-Dk lenses throughout stage 2 to avoid any bias that could arise if subjects were to begin adaptation with their assigned study lenses (i.e., medium- or high-Dk) and then experienced differing group success rates in adaptation. Beginning with 4 hours on the first day, wearing time was increased by 2 hourslday until 8 hourslday was reached. At approximately I-week intervals there- after, subjects reported for follow-up examinations that included history, visual acuity, lens assessment, keratometry, and slit-lamp assessment. Lens changes were made when indicated to help the patient achieve full-time day wear (14-16 hourslday). Additional information on the day wear adaptation phase of this trial has been reported elsewhere.14

On successful adaptation to full-time day wear, the subjects attempted overnight wear and reported in the morning after the first and fourth nights for clinical evaluations. If the subjects had no difficulty with overnight wear (e.g., no discomfort, no red eye) and the clinical assessment.. revealed no complications, a I -week regimen of full-time EW was prescribed consisting of 6 nights of lens wear and 1 night without lens wear. If this EW schedule could be maintained without complications for 4 weeks, the subjects were considered "adapted" and proceeded to stage 3. Subgroups of CLEWS subjects were also asked to participate in a variety of laboratory assessments, including endothelial photomicroscopy and measurements of corneal hydration control, overnight corneal swelling, and epithelial permeability. Baseline laboratory data were taken on participating subjects during stage 2.

Stage 3. Randomized Lens Wear and Follow-up

Subjects entering stage 3 were randomly assigned to either a high- or medium-Dk lens group for 12 months of EW. Clinical assessment examinations were scheduled after randomization at 3, 6, 9, and 12 months. These examinations included an oral history, lens assessment, keratometry. manifest refraction, and a detailed slit- lamp examination. The slit-lamp examination included inspection of lids, conjunctiva, and cornea, with each of these areas further subdivided (e.g., bulbar and palpebral conjunctiva, central and peripheral cornea) for grading by clinicians of 17 possible ocular conditions (e.g., corneal staining, hyperemia, microcysts, infiltrates). Each subdivided area was graded on a 0 to 3 scale (0 = none, 1 = mild, 2 = moderate, and 3 = severe) according to the Mandell Slit Lamp Classification Pro to~ol . '~ Subjects participating in overnight corneal swelling measurements also had pachometry readings taken at 3-month intervals. Subjects who participated in baseline (i.e., prerandomization) endothelial photomicroscopy or corneal hydration control measurements had these procedures repeated at the final 12-month visit. Laboratory results from these subsets of CLEWS subjects will be reported in a separate article.

In addition to the regularly scheduled clinical visits of stage 3, subjects could report for unscheduled visits because of onset of adverse symptoms or at the request of the attending optometrist

Randomization

We assigned subjects to the medium- or high-Dk lens groups by means of a stratified block-randomization scheme. Because age and gender were likely to be associated with outcomes, we used stratified randomization to balance these subject features across the study lens groups. That is, within strata defined by age (530. >30 years) and gender, subjects were assigned to one of the two study lens groups using a randomized block scheme with assignments based on randomly permuted block sizes of two or four subjects. Block randomization ensured that, at most, four consecutive subjects within a stratum could be assigned to the same lens group. Permuting the block sizes (either two or four) ensured that study personnel who were also involved in recruitment activities re- mained masked as to future randomizations in that stratum. Over- all, this randomization strategy ensured balanced allocation of subjects to the two lens groups throughout the course of the trial, thus protecting against possible biases in group comparisons resulting from temporal changes (e.g., altered eligibility criteria, personnel changes, instrument drift, procedural changes, observer criteria changes over time). To implement this scheme, we generated the randomized block assignments for each stratum using random number generating functions in SPlus (Mathsoft, Inc., Seattle, WA).'"hese lists were then stored in SAS (SAS Institute, Inc., Cary, NC)" database tables and accessed by study personnel through an SAS application that included real-time verification of the subject's name, age, gender, and study identification number. ensured that the subject had complete stage 1 and 2 data entered on the study database, and verified that the subject had not already been assigned to a study lens group.

Masking

The PI, optometrists, and technicians were masked to the lens assignment of all patients at all times, as were the patients themselves, which was possible because the two study lenses were indistinguishable by visual inspection. The study coordinator required access to this information to place the lens orders but did not participate in subject examinations or data collection. The biostatistician also had access to this information but retained it in a separate data table that was rarely examined during the trial and was not aware of lens group assignments when making specific decisions about recruitment, retention, or data error resolution. Laboratory measurements were performed by technicians who were not aware of randomization assignments, and optometrists did not have access to laboratory data and referred to prior clinical examination records only when necessary for patient safety. Ex- cept in rare cases, each subject was seen by the same optometrist and technician for each examination throughout the study to control for interobserver variation.

Changes in Protocol from Original Design

Shortly after the start of the study, it became apparent that protocol changes were needed to improve subject recruitment and retention.

Ophthalmology Volume 108, Number 8 , August 2001

After the first 6 months of active recruitment, it was evident that although an adequate number of subjects were attending orientation meetings, more subjects than expected were failing to reach stage 3 randomization. To attract a greater number of potential subjects and to improve subject retention in stages 1 and 2, we implemented four changes in the study protocol: ( I ) reduction of the proposed follow-up period from 3 years to 1 year; (2) relax- ation of the requirement of no prior CL experience to permit some limited history of CL wear; (3) elimination of the arduous corneal hydration control measurements; and (4) reduction of the proposed no-CL wear period at the end of the 12-month follow-up from 3 months to I week. These changes were implemented less than 8 months after the start of recruitment, at which point only 13 subjects had been randomly assigned and only 3 of those had completed the first quarterly clinical assessment. These four changes allowed us to recruit and retain a sufficient number of subjects to complete the clinical trial with adequate statistical power to address our primary aims and without substantially altering the baseline demographic and ocular characteristics of our subject group.

Study Materials, Personnel, and Procedural Review

Lens Materials and Fitting Strategy. RGP lenses were selected for this study, because they are available in a range of oxygen transmissibilities in designs identical in every other respect. The medium- and high-Dk CLs used in this study (paflufocon B and paflufocon D) were made from a Siloxane-fluorocarbon polymer (Paragon Vision Sciences, Mesa, AZ), and both Dk materials have been used extensively for clinical RGP lens wear. The mean Dk of four randomly selected paflufocon B and paflufocon D lenses were 45 X 10-I' and 92 X lo-" (cm2/sec)(ml O,/ml X mmHg), respectively, which are approved by the Food and Drug Admin- istration for EW. On average, these lenses were made with a central thickness of 0.17 mm and varied from 0.12 to 0.22 mm in both Dk groups, which gives a range of oxygen transmissibility (Dkft) of 21 to 38 and 43 to 77 [cm X ml X O,]/[sec X ml X mmHgJ) for the paflufocon B and paflufocon D lenses, respectively. The average overall lens diameter for both lens materials was 9.2 mm. Lenses were fitted with a base curve radius that was slightly longer than the flat corneal meridian as measured by keratometry. For each 1.00 diopter of corneal toxicity (as measured by keratometry), the base curve radius was increased approximately 0.05 mm (i.e., 0.25 diopter). Secondary and peripheral curve radii were made to be 1.0 and 3.0 mm flatter that the central base curve, with widths of 0.2 and 0.3 mm for the secondary and peripheral curves, respectively. Criteria for an acceptable diagnostic fit included either lid attachment (i.e., lens riding slightly under the upper lid after the blink) or interpalpebral fit. In both cases we attempted to achieve optimal lens movement during the interblink period that would allow the lens to gravitate 1 to 2 mm toward the inferior lid margin with a slow vertical movement. An acceptable bearing relationship included an even distribution of fluorescein under the lens and an adequate peripheral lens tear reservoir (without excessive edge standoff).

Study Personnel. Both optometrists and optometric technicians participated in taking study measurements. The optometrists were CL specialists with considerable clinical experience in RGP design, fitting, and follow-up care, who were responsible for performing the initial eye examinations, CL fittings, epithelial permeability measurements, and all clinical examinations during the 12 months of EW. Optometric technicians were trained to perform pachometry, endothelial photomicroscopy and digitizing, lens verification, and subject training in CL handling procedures.

All technicians successfully completed a rigorous certification process for each instrument and laboratory procedure and were periodically retested to ensure that they maintained a consistently high standard. In addition to the optometrists and technicians, the research team included a faculty biostatistician who, as co-PI for most of the CLEWS study, was instrumental in the design and implementation of the trial including review by a DSMB and the creation of data analysis strategies. A senior biostatistician was responsible for monitoring of recruitment and retention, database design and data management, and working with the faculty biostatistician and the PI in statistical analyses. A study coordinator was responsible for patient scheduling, implementing recruitment procedures (e.g., posting advertisements, contacting the clinic for information on referrals), organizing and filing all patient records, and some data entry and computer file management under the supervision of the biostatistician.

External Interim Review. CLEWS received institutional ap- proval from the University of California at Berkeley Committee for Protection of Human Subjects. In addition, an independent DSMB was appointed to conduct an interim review of CLEWS with regard to safety and related issues. Members of the DSMB consisted of ( I) an infectious disease physician/epidemiologist, (2) a biostatistician (chairperson), (3) a university-based optometrist with expertise in CL fitting, (4) an ophthalmologist with expertise in external disease, and (5) a community member. When approximately 60% of the study was complete, the DSMB received presentations concerning implementation of the study design, efforts to protect human subjects, and preliminary data related to safety. The DSMB concluded that the study was being conducted in a manner consistent with providing optimal protection to the subjects and that the study lenses had caused no adverse ocular events of a severity to warrant discontinuation of the study. The DSMB concluded that the alterations to the study design in response to recruiting and retention problems were timely and 0th- - erwise appropriate and that the study retained sufficient statistical power to achieve its primary aims. The committee recommended that the trial continue and that an additional meeting was not necessary unless new safety issues arose.

Data Management and Quality Assurance

The number and variety of measurements and observations performed under this study design generated a vast array of demographic, historical, clinical, and laboratory data. It was necessary to develop a method of data entry and verification that would permit their timely review for monitoring recruitment and retention rates and for assuring patient safety and high data quality. We used the Teleform system (Cardiff Software, San Marcos, CA),'" which uses optical mark and character recognition for scanning data entry forms, provides tools for immediate verification of accuracy, and permits design of customized data entry forms that the optometrists could easily complete without disrupting the flow of the examination or rapport with the patient. The study coordinator was responsible for scanning forms, verifying data, and working with the clinicians and biostatisticians to resolve data entry errors. Details of the form design, data validation, and performance capabilities of the Teleform system have been reported elsewhere. l y

After scanning and verification, the data were subjected to an extensive set of quality assurance (QA) programs written for the SAS system. The CLEWS QA system consisted of three levels of SAS programs: (I) within-form checks for missing data, valid data ranges and types, and logical relationships among form variables; (2) comparison of new batches of scanned data to the master CLEWS database, checks for duplicate forms, logical relationships . between current and prior patient data, and errors in the sequence

Fusaro et a1 Berkeley Contact Lens Extended Wear S t u d y : D e s i g n

OrientationlScreening ~~~ ~~e-l-, l3e.I I Fail 1

1241 432

1673 Stage 1 Failures

+ 545 Pass

545 Enter * DW Adaptation 1-bp- EW Adaptation

Stage 2 (Endo Photos. OIN Swell, CHC)

Fail Fail Pass

194 150 t

344 Stage 2 Failures 201 Pass I

72 Stage 3 Terminations

Figure 1. At any stage of the Contact Lens Extended Wear Study, clinical trial subjects could elcct to discontinue participation, be dropped hy the clinician, or successfully complete the stage and proceed to the next. Shown within each of the three stages are the numhcr of subjects entering rhar stagc, the various examinations and measurements performed, the number of subjects failing to complete these visits, and the number successfi~lly completing all visits for that stage. Thc timeline at the bottom shows when data collection for each stage hegan and was completed ovcr the course of the 5-year study. (CHC = corneal hydration control asscssrnents; Clin Examination = quarterly clinical follow-up examinations; Endo photos = endothelial photomi. crographs; O/N Swell = overnight corneal swelling measurements.) DW = day wear; EW = extended wear.

of visit dates; and (3) screening of the master database for logical inconsistencies between different variables across different visits. All laboratory data were also subjected to customized SAS QA programs.

The QA system generated reports that flagged discrepancies in the database that could represent either data entry errors or real changes in the condition of the patient requiring an investigation by the PI or attending optometrists (e.g., large changes in visual acuity). The QA reports were first screened by the biostatistician, and any data entry errors were resolved by referring to the data entry forms and, if necessary, consulting the study coordinator or optometrist. Flagged items that were not simple data entry errors could possibly indicate real changes in patient status and thus were referred to the PI for investigation. Data entry errors

were more common by far than real changes, although we did occasionally encounter large changes in corneal curvature (e.g., > I .O diopter) or reductions in visual acuity (e.g., > 2 lines on the visual acuity chart) that prompted optometrist review of the patient records and scheduling of the patient for follow-up examinations. The QA reports did not reveal any previously unidentified adverse events severe enough to warrant discontinuation of a subject from the study. All CLEWS data, clinical records, laboratory readings, and QA reports were backed up at the instrument or scanning site, and multiple copies were stored outside the laboratory area. New data were backed up fre- quently as forms were scanned, QA reports completed, or instrument readings performed, whereas the master database was backed up monthly.


Stage 1 Failures Stage 2 Failures \

Discomfort at Clinician decision, 100% 1 3 2 % initial lens fitting 13.08% not related to lens wear

39,57% General disinterest after 17.15% Clinician decision,

orientation and screening related to lens wear

14.1 1% Prior history of CL wear Subject decision, 33'72% related to lens wear

Failed to meet Subject decision, 45.01% other eligibility criteria 36.05% not related to lens wear

2218 Entered 1673 Failed to Proceed

545 Entered 344 Failed to Proceed

Figure 2. During stage 1 of the Contact Lens Extended Wear Study tr~al, subjects could discontinue participation either because of discomfort at the initial fitting or hecause of factors unrelated to contact lens (CL) wear (e.g., time commitment, moving residence out of the area), or study personnel determining that an interested pcltcntial suhjecr did not meet eligibility criteria (approximately 14% for prior CL wear). Subjects entering stage 2 could elect to discontinue because of discomfort cluring adaptation to daily or overnight wear or for reasons unrelated to lens wear, whereas the clinician could drop a suhject because of lens-rclaretl complications (approximately 17%) or noncompliance with the srully protocol.

Results

S u b j e c t Recruitment and Retention

Figure 1 depicts the flow of subjects through the three stages of the trial from the beginning of recruitment to the end of data collection and termination from the study. A total of 2218 subjects respond- ing to recruitment efforts entered stage 1 by attending a CLEWS orientation from April 1994 through December 1997. Of these 2218 prospective subjects, 977 met the initial eligibility criteria and were scheduled for comprehensive eye and CL fitting examinations. Of these 977 subjects, 545 subjects met all the requirements for CL wear and study eligibility and were dispensed high-Dk RGP adaptation lenses for entry into stage 2. Figure 2 shows the percentages of subjects who failed in either stage 1 or stage 2. Of the 2218 subjects who entered stage I , a total of 1673 did not proceed to stage 2 (75%). the primary reasons being that they failed to meet eligibility criteria (59% of failures) or ex- pressed a lack of interest in participating in CLEWS after the study requirements were fully explained (40%). Among the 545 who entered stage 2, 344 failed to reach stage 3 randomization (63%). Of those 344 subjects 17% had CL-related responses that were sufficiently important to discontinue lens wear (typically central corneal staining or persistent lens adherence), whereas another 34% of withdrawals were subject-initiated, primarily because of lens-associated discomfort. For the remaining subjects (49%) who did not proceed to stage 3, approximately 13% were dropped by the clinician because of noncompliance with the CL adaptation protocol, whereas 36% were due to issues not related to CL wear, such as moving out of the area, difficulty in maintaining a commitment to the study visit schedule, or nonocular related health issues.

Application of Hypoxic Dosage

The two study groups were allocated lenses with substantially different Dk. If all other lens parameters were kept constant, the two groups would have received two different levels of hypoxia

corresponding to the different lens Dk. However, the amount of oxygen reaching the cornea is determined by the oxygen transmissibility (DWt), which is a function of both the permeability of the material and the thickness of the lens. Because lens thickness varies with vertex power and study lenses varied in power among CLEWS subjects, there was a range of Dklt values for subjects . within each Dk group. To verify that the medium- and high-Dk groups were subject to distinctly different hypoxic doses (i.e., different levels of oxygen reaching the cornea, determined by the Dklt), we examined the distribution of Dk/t when lens thickness is taken into account for each subject. The histogram of Dklt shown in Figure 3 reveals that, despite the variability within each Dk lens group caused by the subject-specific lens thicknesses, the distributions of oxygen transmissibility are completely disjoint between the two groups. We therefore conclude that the study groups did receive different hypoxic doses.

I n t e r n a l and External Validity

Of the 201 subjects randomly assigned to study lenses, 103 subjects received the medium-Dk lenses and 98 received the high-Dk lenses. Table 2 lists the baseline (i.e., prefitting) characteristics of the two study groups and shows that the randomization scheme maintained balance between the two groups in age and gender. There were no important differences between the lens groups in mean age, gender or ethnic composition, proportion with prior CL history, mean corneal curvature, refractive error, or visual acuity. The close similarity of these groups in terms of their baseline characteristics ensures the internal validity of between-lens group comparisons (i.e., because the two groups are comparable with respect to the characteristics listed in Table 2, any apparent differences we observe between groups in study outcomes will not be due to confounding by these variables).

It is also important to compare the baseline characteristics of the 344 subjects failing stage 2 to those of the 201 subjects who entered stage 3 and whose data will be used in subsequent analy- - ses, because differences between randomly assigned subjects and

Fusaro et al Berkeley Contact Lens Extended Wear Study: Design

I

Medium-Dk ; : High-Dk I

Min D M Max DM I Min Dkll = 21 = 38 j = 4 3

I

Max DM = 77

Mean oxygen transmissibility [lo-' (cm x mL x O,)/(sec x mL x mm Hg)]

Figure 3. Although the two randomized study groups received lenses of different oxygen permeabilities (Dk), it was possible that the range of oxygen transmissibilities (Dk/t) caused by differences in lens thickness within each group could result in overlap in the hypoxic doses that the two groups received. The histogram shows that the distributions of Dkh for the two groups were separated, confirming that distinctly different hypoxic doses were delivered to the study groups.

Table 2. Baseline Characteristics of Subjects who met Contact Lens Extended Wear Study Eligibility Criteria, Successfully

Adapted to Rigid Gas-Permeable Extended Wear, and Were Randomly Assigned to Either the Medium- or High-Oxygen

Transmissibility Group. The Similarity of These Characteristics in the Two Study Groups Ensures that Apparent Group

Differences in Outcomes Will Not Be the Result of Confounding by These Variables

Medium-Oxygen High-Oxygen Permeability Permeability (n = 103) (n = 98)

Subjects with contact lens history (%) Gender (%)

Male Female

Ethnicity (%) African American Asian American Caucasian Hispanic American Native American Other

Mean age at orientation (years) Mean corneal curvature (diopters)

Horizontal meridian Vertical meridian Toricity Residual astigmatism

Mean refractive error (diopter) Sphere Cylinder Equivalent sphere

w Mean visual acuity log(l/Snellen fraction)

Table 3. Baseline Characteristics of Subjects Who Were Randomly Assigned vs. Those Who Began Stage 2 Lens Wear

Adaptation but Failed to Proceed to Randomization. The Similarities between Eligible Subjects from Whom Data Were or Were Not Obtained Suggests that Statistical Inference from

our Stage 3 Results to the General Contact Lens Extended Wear Study Population is Appropriate

Randomly Failed Stage 2 Assigned

Adaptation to Stage 3 (n = 344) (n = 201)

Subjects with concact Lens history (%) 23.26 20.40 Gender (%)

Male 58.14 56.22 Female 41.86 43.78

Ethnicity (%) African American 7.56 5.97 Asian American 41.57 41.29 Caucasian 3 1.69 37.31 Hispanic American 10.47 7.46 Native American 0.29 1 .OO Other 8.43 6.97

Mean age at orientation (years) 23.88 23.09 Mean corneal curvature (diopters)

Horizontal meridian 43.20 43.48 Vertical meridian 43.82 44.15 Toricity 0.75 0.77 Residual astigmatism 1.15 1.25

Mean refractive error (diopters) Sphere -2.52 -2.81 Cylinder -0.41 -0.48 Equivalent sphere -2.73 -3.06

Mean visual acuity log(l/Snellen fraction) 0.97 1.01


the original pool of eligible subjects might limit the external ReferenceS validity, or "generalizability," of our statistical results. That is, if the subjects who successfully adapted to RGP EW and entered stage 3 had different demographic or ocular features from the eligible candidates who were not able to adapt to EW, statistical results obtained using this sample would not necessarily be appli- cable to the general CLEWS study population as defined by the entrance criteria, adaptation criteria, and recruitment scheme. Ta- ble 3 shows the baseline characteristics of the 344 stage 2 failures and the 201 stage 3 randomly assigned subjects. As expected from the pool of potential subjects from the Berkeley campus and surrounding community, the mean age of initial entrants was roughly 23 years, approximately 85% of subjects were students (not shown in Table 3), and Asian-Americans and Caucasians accounted for approximately 75% of the ethnic makeup. Mean visual acuity, male-to-female ratios, and proportions of subjects with prior CL history were similar. Although there were very modest differences in mean corneal curvature and refractive error, such differences are insignificant clinically and within the usual measurement accuracy for these instruments. These comparisons show that the eligible candidates who were unable to complete EW adaptation did not display any clinically important differences compared with the sample of 201 subjects who we randomly assigned and for whom we collected follow-up data.

Conclusions

In this article we have described the purpose, design, and conduct of the CLEWS clinical trial, the nature of the hypoxic treatment, the protocols used to control bias and ensure data integrity, and the baseline characteristics of the CLEWS study population.

In summary, CLEWS was appropriately designed to address the study hypotheses and was conducted with regard for the safety of the subjects and the integrity of study data. CLEWS adhered to rigorous protocols that were successful in controlling for bias and maintaining balance in the two study groups with respect to potential confounding variables. Finally, we have provided the information on our hypoxic treatment and the characteristics of our subjects necessary for evaluating and interpreting the results that follow in the companion article.

Acknowledgments. We wish to acknowledge the contribu- tions of the members of the Data and Safety Monitoring Board (Joan Hilton, ScD, DSMB, Committee Chair, Department of Ep- idemiology and Biostatistics, University of California San Fran- cisco; David Heiden, MD, School of Optometry, University of California Berkeley; Arthur L. Reingold, MD, School of Public Health, University of California Berkeley; Barry Weissman, OD, PhD. Department of Ophthalmology, University of California Los Angeles School of Medicine; and Barbara Woodruff, Chief Ad- ministrative Officer, School of Optometry, University of Califor- nia Berkeley. We also wish to thank John Fiorillo, who provided technical assistance in the preparation of this article.

Poggio EC, Glynn RJ, Schein OD, et al. The incidence of ulcerative keratitis among users of daily-wear and extended- wear soft contact lenses. N Engl J Med 1989;321:779-83. Schein OD, Glynn RJ, Poggio EC, et al. The relative risk of ulcerative keratitis among users of daily-wear and extended- wear soft contact lenses. A case-control study. Microbial Keratitis Study Group. N Engl J Med 1989;321:773-8. Kaufman HE. Problems associated with prolonged wear soft contact lenses. Ophthalmology 1979;86:411-7. Holden BA, Sweeney DF, Vannas A, et al. Effects of long- term extended contact lens wear on the human cornea. Invest Ophthalmol Vis Sci 1985;26: 1489-501. Sankaridurg PR, Sweeney DF, Sharma S, et al. Adverse events with extended wear of disposable hydrogels: results for the first 13 months of lens wear. Ophthalmology 1999; 106: 1671- 0 -

8U. 6. Mac Rae SM, Matsuda M, Shellans S, Rich LF. The effects of

hard and soft contact lenses on the corneal endothelium. Am J Ophthalmol 1986; 102:50-7.

7. Polse KA, Decker M. Oxygen tension under a contact lens. Invest Ophthalmol Vis Sci 1979;18: 188-93.

8. Holden BA, Mertz GW. Critical oxygen levels to avoid corneal edema for daily and extended wear contact lenses. Invest Ophthalmol Vis Sci 1984;25: 1161-7.

9. Ren DH, Petrol1 WM, Jester JV, et al. The relationship between contact lens oxygen permeability and binding of Pseudomanas aeruginosa to human corneal epithelial cells after overnight and extended wear [published erratum appears in CLAO J 1999;25:175]. CLAO J 1999;25:80-100.

10. Polse KA, Brand RJ, Cohen SR, Guillon M. Hypoxic effects on corneal morphology and function. Invest Ophthalmol Vis . Sci 1990;31: 1542-54.

11. Bonanno JA, Polse KA. Corneal acidosis during contact lens wear: effects of hypoxia and CO,. Invest Ophthalmol Vis Sci 1987;28:1514-20.

12. Cohen SR, Polse KA, Brand RJ, Bonanno JA. Stromal acidosis affects corneal hydration control. Invest Ophthalmol Vis Sci 1992;33: 134-42.

13. Guillon M. Basic contact lens fitting. In: Ruben M, Guillon M, eds. Contact Lens Practice. New York: Chapman & Hall Medical, 1994; chap. 27;602-22.

14. Polse KA, Graham AD, Fusaro RE, et al. Predicting RGP daily wear success. CLAO J 1999;25: 152-8.

15. Mandell RB. Slit lamp classification system. J Am Optom ASSOC 1987;58: 198-201.

16. Mathsoft. S-PLUS User Guide, Version 4.0, Seattle: Math- Soft, Inc, 1997.

17. SAS Institute Inc. SAS Language. Reference, Version 6,lst ed. Cary, NC: SAS Institute Inc, 1990.

18. Cardiff Software. Teleform User Guide, Version 5. San Mar- cos, CA. Cardiff Software, Inc, 1996.

19. Graham AD, Fusaro RE, Polse KA. Automating Data Entry and Validation in Clinical Research. Berkeley: Group in Biostatistics, School of Public Health, University of Cali- fornia at Berkeley. 1998; Biostatistics Technical Reports #71.

To order reprints, call: 1-212-633-3813: fax: 1-212-633-3820 email: [email protected]

. -. The Berkeley Contact Lens Extended Wear Study: Part I1 Clinical Results

KenGeth A. Polse, OD, MS, ' Andrew D. Graham, MS,' Robert E. Fusaro, PhD, ID,' Cheslyn M. Gan, OD, ' Reuben K. Rivera, OD, MS,' Meng C. Lin, OD,' Timothy L. Sanders, OD,' Nancy A. McNamara, OD, PhD,' Judy S. Chan, OD, MS'

Objective: To describe the principal clinical outcomes associated with 12 months use of rigid gas-permeable (RGP) extended wear contact lenses and address two primary study questions: (1) does extended wear (EW) of high oxygen transmissibility (Dklt) RGP lenses reduce the incidence of ocular complications, and (2) does the wearing of high-Dk/t lenses reduce the rate of failure to maintain 6-night RGPEW over 12 months?

Design: A randomized, concurrently controlled clinical trial. Intervention: Subjects who adapted to EW with high Dk (oxygen permeability) RGP lenses were randomized

to either high Dk or medium-Dk RGP lenses for 12 months of 6-night EW: Main Outcome Measures: Contact lens-associated keratopathies (CLAK), changes in refractive error and

corneal curvature, and survival in EW. Results: Two hundred one subjects were randomized to medium' or high-Dk lenses for 12 months of EW.

Sixty-two percent of the subjects in each group completed 12 months of EW; however, the probability of failure was significantly greater for the medium-Dk group. Although the risk of complications was similar for the two groups, the number of CLAK events that led to termination were 16 versus 5 for the medium-Dk and high-Dk groups, respectively. This suggests that the type of adverse response or the inability to reverse an adverse event was different for the group being exposed to the lower oxygen dose.

Conclusions: The level of oxygen available to the cornea has a significant impact on maintainir successful RGP extended contact lens wear, but not on the initial onset of CLAK. The number of c l i n i c w events leading to termination was substantially higher for the medium Dk group, which suggests that corneal hypoxia is an important factor in the development of CLAK. Although overnight contact lens wear should be recommended with caution and carefully monitored for early detection of ocular complications, it appears that high-Dk RGP lenses can be a safe and effective treatment for correction of refractive error for most individuals who can adapt to EW. Ophthalmology 2001;108:1389-1399 O 2001 by the American Academy of Ophthalmology.

The Berkeley Contact Lens Extended Wear Study (CLEWS) was a randomized, concurrently controlled clinical trial designed to examine the effects of corneal hypoxia on ocular complications arising from the extended wear of rigid gas-permeable (RGP) contact lenses of either medium or high oxygen permeability (Dk). In part I of this report, we described the background, design.

Originally received: August 25. 2000. Accepted: klarch 6. 2001. Manuscript no. 200499.

' Morton D. Sarver Laboratory for Cornea and Contact Lens Research. School of Optometry. University of California. Berkeley. California. ' Division OF Public Health Biology and Epidemiology. School of Public Health, University of California. ~ c r k e l e ~ . - ~ a l i f o m i ~ . - Supported by the National Eye Institute. National Institutes of Health. Brthesda. Mary1;lnd (grant no.: EY07728); Concise Contact Lens Co.. San Lenndro. California: and Paragon Vision Sciences. Mesa. Arizona. The authors have no proprietary interests in this article. Reprint requests to Kenneth A. Polse. OD. MS. School of Optometry. University of Colilbrnia. Berkeley. CA L94720-203-0.

and implementation of CLEWS. In that article, we showed that the distributions of lens oxygen transmissibility (Dklt) in the lenses assigned to the two study groups were completely disjoint (Fusaro RE, Polse KA, Graham AD? The Berkeley Contact Lens Extended Wear Study. Part [: Study Design and Conduct), confirming that subjects randomized to the inedium-Dk lens group were subject to a greater degree of hypoxia. Providing different oxygen levels to each group permitted us to explore the association between hypoxic dose and certain clinical outcomes associated with 12 months of rigid gas-permeable extended wear (RGPEW) by addressing two primary study questions:

I . Does extended wear of high-Dk RGP lenses, rather than medium-Dk RGP lenses, reduce the incidence of ocular complications?

2. Does the wearing of high-Dk RGP lenses, rather than medium-Dk RGP lenses, reduce the rate of failure maintain 6-night RGPEW? 4

O ?OUI by the American Aci~drmy of Oph~hal~nology

Published hy Elsevier Scicocr Inu. ISSN 0161-MZOIOllf-see front matter 1389

PI1 SOI61-bJ?OtOl )(WK?Y-.i


Materials and Methods

* In this section. we briefly review the CLEWS design, explain our strategy for quantifying various types of ocular complications for analysis, and present the statistical methodology we used in ana- lyzing the CLEWS clinical trial data. The reader is encouraged to refer to part I for a more comprehensive treatment of the issues relevant to assessing the integrity of the data and to interpreting and generalizing the results presented in this companion article.

Overview of the Berkeley Contact Lens Extended Wear Study Design

The Berkeley CLEWS was a randomized, concurrently controlled clinical trial designed to evaluate the hypothesis that ocular mor- bidity in RGPEW is associated with the level of corneal hypoxia induced by the lenses. The trial was structured with three main stages. In stage I . prospective subjects were oriented to the goals and procedures of the study qnd then examined to determine eligibility. In stage 2, subjects were adapted to contact lens wear, beginning with daily wear and progressing to full-time (i.e., 6 nights per week) extended wear (EW). Those who were able to achieve full-time EW then proceeded to stage 3 and were randomly assigned either medium- or high-Dk lenses for 12 months of full-time EW. Stage 3 subjects attended quarterly visits, during which optometrists conducted detailed interviews. slit-lamp examinations, visual acuity assessments, and refractive error (RE) and corneal curvature (K) readings. Results of stage I and stage 2 recruitment and retention efforts .and baseline subject data are presented in part I. The clinical outcomes for the two randomized subject groups are compared in this article.

Contact Lens-associated Keratopathy

To compare the incidence of ocular complications between the two lens groups, it was necessary to define contact lens-associated keratopathy (CLAK). We asked six experienced contact lens prac- titioners who were not associated with CLEWS and who were unaware of any CLEWS interim results to complete a survey detailing the severity of complications that they believed would typically require intervention (e.g., changing lens parameters, pre- scribing medications, temporarily discontinuing lens wear to allow the complication to subside, completely tenninating lens wear). The survey respondents were presented with 17 possible keratopathies and asked to indicate for each condition the minimum g a d e of severity (0 = none, 1 = mild, 2 = moderate, 3 = severe) that they would consider an adverse event requiring clinician intervention. Each clinician was also asked to indicate the number of events of lesser severity that, if observed concurrently, would require intervention. Finally, they were asked to indicate the diopters of refractive error (RE) and corneal curvature (K) change from prefitting measurements that they would characterize as an adverse response requiring intervention. The survey results led to the CLAK definitions reported in Table I.

Our goal was to quantify CLAK in a manner that would reflect the consensus opinion of the outside clinicians and be meaningful in practice (i.e., as adverse responses severe enough to require clinician intervention). Using the survey results. we constructed four increasingly inclusive definitions of CLAK and repeated our analyses using each of these four definitions. With this approach, we were able to assess the ocular response to hypoxic dose while allowing for more or less restrictive definitions of CLAK. thereby facilitating comparison with other studies that may use different criteria in reporting incidences of lens-related complications. Us- ing our first definition of CLAK (type I). we examined the effects

Table I. Definitions of Contact Lens-associated Keratopathy

Minimum Value Defined as Contact

Kentopathy or Lens-associated Adverse Event Kentopathy

Bulbar conjucrivn Injection 7

Stilining Superior I Inferitlr 1

N;~siil 7

Tctnpral 2 Pi~lpehral cunjuncrtval injectiiln

Superior tarsus 2 Inferior tilrsus 'J

Hypertrophy Superior tarsus 2 Inferior tarsus 1

Corneal staining Cenrrnl I Peripheral 2

Corneal inhlrrates Central I Peripheral I

Microcysts 2 Srriae 2 Polymegethism 2 Tear debris 2 Lens adherence 3 times weekly, persistent

Change in corneal curvature Horizontal meridian 0.88 Vrrrical meridian 0.S3

Change in refractive error Sphere 0.88 Cylinder 0.92

Simultaneous subclinical slit-lamp 3 findings (n)

Discomfort and related symptoms 2

Type 1 contact lens.associated keratopathy includes slit-lamp hndings and lens adherence; type 2 includes both type I responses and changes in corneal Gwature or refractive error; type 3 includes both type 2 and multiple, simultaneous subclinical hndings; type 4 includes both type 3 and subjective symptoms related to discomfort. Severity grading: 0 = none. 1 = mild, 2 = moderate. 3 = severe.

of hypoxic dose on the occurrence of any keratopathy detected by slit-lamp examination in either eye, whose severity equaled or exceeded the average of the six survey responses rounded to the nearest severity grade. Type 1 CLAK also included lens adherence that required changes to the lens parameters. Lens binding that was less frequent or was alleviated by the use of solutions, cleaning or polishing the lens, or further training of the subject in lens care and handling procedures was not considered a type 1 CLAK. Lens adherence was not included in the original survey because it was thought that individuals subject to persistent lens adherence would not be able to complete the EW adaptation period, and therefore adherence would not occur in our randomized subjects. However. a data quality review revealed that several randomized subjects did experience persistent lens binding. thus prompting a manual search of the entire bank of patient records to ascertain all such cases of adherence. Randomization assignment was not revealed during the search, and at the time the decision was made to include lens adherence as a type I CLAK. i t was not known whether the two Dk groups differed in adherence rate.

We then eipanded our definition of CLAK (type 2) to include type I CLAK and changes in K or RE from baseline in either eye

Poke et a1 . Berkeley Contact Lens Extended Wear Study: Results

that exceeded the mean diopters of change indicated by the survey respondents. Our third definition of CLAK (type 3) included three or more simultaneous "subclinical keratopathies" (i.e.. observed at the slit lamp, but less severe than the grades qualifying as type I) in either eye, in addition to type 2 CLAK. Finally. we repeated our analyses using a fourth definition (type 4) that included type 3 CLAK or moderate to severe discomfort, burning, itching, or pain at a regularly scheduled clinical assessment, or these symptoms to such a degree that an unscheduled visit was needed.

Statistical Methods

An attractively simple way to summarize the failure experience of subjects at risk for an adverse event is by the simple proportion of subjects who fail over the course of the study. This approach is inappropriate, however, For several reasons. First, it is inefficient in that i t fails to use data concerning the specific times at which failures occurred and. rather, treats all failures within the study period as the same. Second, this approach ignores the fact that only incomplete, or censored. information may be available for some subjects. Subjects may provide partial, right-censored data because they elected to discontinue study participation or were terminated from the study by the clinician for reasons unrelated to study lens wear, such as time commitment, general disinterest, or noncompliance with the study protocol. An unknown number of these subjects may have suffered lens-related adverse events had they been available for the entire 12-month observation period. Subjects were failed for noncompliance by the clinician if they were unable or unwilling to attend the required laboratory visits consistently, if there were frequent lens care or handling problems as a result of failure to follow instructions or if subjects were inconsistent in maintaining their 6-night EW schedule. For subjects who missed scheduled visits and did not contact the laboratory. the Study Coordinator followed a standard procedure for attempting to contact the subject before completing a termination form. Adverse events were defined as initial onset of a CLAK condition (for study question I) and termination from the study as a result of a type 4 CLAK condition (for study question 2).

To account for censoring, we used the product-limit (Kaplan- Meier) approach' to estimate the Dk group-specific survival curves, which characterize the probabilities of successful EW or freedom from complications over time. The homogeneity of the survival curves between Dk groups was assessed using a log-rank test.' Estimates of the relative risk of complications or RGPEW failure associated with hypoxic dose were obtained by fitting Cox proportional hazards models.' We used the SAS (SAS Institute, Inc., Cary. NC) procedures LIFETEST and PHREG to implement these survival analytic techniques.'

It has been suggested that overnight lens wear, particularly with lower-Dk materials, may alter corneal topography.'.' We therefore attempted to ascertain whether K or RE in our RGPEW subjects differed between Dk groups or changed over time in either group. We used the SAS procedure MIXED to implement a mixed effects analysis of variance approach to comparing changes from baseline in K (horizontal and vertical meridians) and RE (sphere and cylinder components) in the two Dk groups.' We modeled the change from baseline in each of these outcomes with fixed effects for Dk group and visit (3.6.9. and 12 months) within each group and a compound symmetric covariance structure to account for thk correlations among repeated measurements on the two eyes of each subject across visits. This structure assumes a common covariance among all measurements on a single subject.

Results i

Overview J

This section is divided into four parts. In the first part. we describe subject participation rates by visit along with the reasons subjects dropped out or were terminated from the study. The second section presents our basic clinical observations, including group trends in RE and K, the specific types of keratopathy encountered. and detailed descriptions of those subjects in each group who had to discontinue overnight lens wear completely because of CLAK. The third section addresses the first of our primary CLEWS questions by comparing the probabilities of the two types of lens wearers remaining free from complications through time and by estimating the relative risk of complications associated with hypoxic dose. Finally. in the fourth section. we address the second of our two primary CLEWS questions by comparing the estimated survival probability curves between groups and estimating the relative risk of failure to maintain EW associated with hypoxic dose.

Subject Retention, Termination, and Loss to Follow-up . Of 2218 subjects who attended an orientation meeting, 545 subjects were deemed eligible and were willing to participate in CLEWS. These subjects were fit with high-Dk lenses for daily wear adaptation. Of the 344 subjects who proceeded from daily wear to EW adaptation. 201 subjects adapted to EW and were randomized to either medium-Dk (n = 103) or high-Dk (n = 98) RGP lenses for 12 months of full-time EW. Our stratified block randomization scheme. detailed in part I of this report, ensured thp1 approximately the same number of subjects were assigned to er lens group at any given point during the study and that ages a* genders were balanced between the two groups.

Table 2 shows the number of subjects examined at each quarterly visit in the two goups and the number failing to attend each scheduled assessment. Of the 103 medium-Dk wearers. 24 subjects did not complete the first quarterly visit, and data were collected on the remaining 79 subjects. Of these 24 subjects who did not participate in the 3-month examination, 3 returned for subsequent visits, whereas 21 were either lost to follow-up or failed as a result of complications before the first quarterly visit. The high-Dk group showed a similar pattern; 19 of the 98 randomized subjects failed to complete the first visit, leaving 79 available for 3-month observations. After these relatively large losses of subjects early in the course of follow-up. the number of subjects failing to complete subsequent visits declined. At final count, 64 subjects (62.1%) in the medium-Dk group and 61 subjects (62.2%) in the high-Dk group were able to complete the entire 12 months of RGPEW successfully.

During the postrandomization phase, some subjects required one or more appointments in addition to the quarterly assessments. These .unscheduled visits were either initiated by the patient because of a lens-related problem or an ocular symptom or by the clinician at the time of a regularly scheduled clinical assessment as a result of a slit-lamp finding requiring monitoring or follow-up care. The reasons reported for unscheduled visits are shown in Table 3. Over the postrandomization period, there were a total of 91 unscheduled visits from 53 subjects in the medium-Dk group and 77 visits from 42 subjects in the high-Dk group. Replacement of lost lenses was the most common reason for unscheduled visits. although there were a substantial number of visits as a result

'

onset of adverse symptoms, primarily red eye and foreign b o w sensation in both groups, discomfort and blurred vision in the medium-Dk group, and pain in the high-Dk group. The most

Ophthalmology Volume 108, Number 8 , August 2001

Table 2. Contact Lens-associated Kentopathy CLEWS Subject Participation

Medium Oxygen permeability Group High Oxygen Permeability Group - ( n = 103) (n = 98)

Follow-up Visit (mos) 3 6 9 I2 3 6 9 I 2

Completed visit 79 7 3 66 64 79 67 64 61 Did not coniplete visit 2 I 7 7 2 IS 1 1 4 2

Cirrnul;~rivr !I 28 3 5 37 I Y 29 33 35 Missell visit but rerurncJ J 2 2 2 1 2 I 2

Shown ore the numhers OF contact lens-associi~teJ keratopethy sirbjrcts in cacll c~ndo~nined sruJy proup who complrred or failcrl ro cornplece each t schdulcd it)llow-up visit. Occasionally, subjects woul~l miss :I scheduled visit without discontinuing srtdy plrriciparion (e.g.. for on extended vacation)

unrl rcrum to the stuily for subscqi~ent visirs.

common slit-lamp findings at the unscheduled visits were corneal and bulbar conjunctival staining, with few or no occurrences of corneal striae. hypertrophy. infiltrates, microcysts, injection, or polymegethism.

The reasons for termination or drop-out before 12 months in the two study groups are provided in Figure 1. Most failures and dropouts in both groups occurred within 3 months after randomization. although the reasons for failure differ between the groups. In the high-Dk group, most subjects unavailable for the first quarterly visit (n = 18) either elected to drop out for reasons unrelated to contact lens wear or were terminated for noncompliance, whereas only two subjects failed for lens-related adverse responses. In contrast, in the medium-Dk group. the reasons for failures before the first quarterly visit were nearly evenly divided between those unrelated to lens wear (n = I I) and lens-related adverse responses (n = 10). Over the remaining visits, the medium-Dk group continued to demonstrate comparable numbers of

v failures for lens-related complications (n = 7) and reasons unrelated to contact lens wear (n = 8). The high-Dk group also continued its pattern of reasons For failure, with far more unrelated to lens wear (n = 14) than lens-related adverse responses (n = 2).

Table 3. Reasons for Unscheduled Visits for Initial Onset of Symptoms (Follow-up Examinations for Previously Identified

Conditions Not Included in This Summary)

Medium Oxygen High Oxygen Permeability Permeability

Group Group

No. subjecrs requiring unscheduled visits

Total number of unscheduled visits by these subjects

No. subjects reporting reason for visit (may be > I reason. > 1 visit per subject, or both)

Losr lens Red eye Pain Discomfort Foreign M y sensation Blurred vision Tearing Drying Burning Itching Photophobia Crackedbroken lens Discharge Lens handling problem

Clinical Observations In this section. we present further observational data that do not. by themselves, directly address the two primary CLEWS research questions, but nevertheless are of clinical interest. We first report on changes observed in K and RE, then describe the specific types of keratopathy encountered in the two lens groups and present the detailed diagnoses of the 21 subjects failing for lens-related ocular complications. In the two sections that follow this clinically oriented presentation, we apply the appropriate survival analytic techniques to these data to address directly the two primary questions for which CLEWS was designed.

Changes in Refractive Error and Corneal Curvature. To determine whether overnight wear of medium- or high-Dk RCP lenses is associated with changes in K or RE over time, manifest refraction and keratometry readings were performed at the baseline subject evaluation and at each of the quarterly clinical assessments. For both Dk groups, average changes in the spherical refractive component from baseline measurements were small and in the direction of reduced myopia (ranging from 0.12-0.20 diopters [Dl at the different visits), and for the cylinder there were small reductions in the astigmatic component (ranging from 0.1 1-0.15 D). Keratometry for both Dk groups showed modest decreases in power in both the vertical and horizontal meridians (ranging from 0.46-0.5 1 D). There were no significant Dk group or visit effects on any of these measures within the two Dk groups. In Figure 2, the change from baseline in refractive sphere, using the average of each subject's two eyes, is shown stratified by visit for the medium- and high-Dk groups, illustrating the lack of clinically important changes in corneal topography found in both RE and keratometry measurements. From these results, we concluded that overnight wear of RGP lenses does not significantly affect corneal topography, nor does the oxygen permeability of the lens.

0bservations.of Lens-related Complications. Table 4 shows the overall number of cases of CLAK observed during the CLEWS clinical trial for each of our definitions, under all four of which a slightly larger number of subjects had complications in the medium-Dk group than in the high-Dk group. Of potential clinical interest are the specific types of slit-lamp findings from up to I year of observation of more than 200 RGPEW patients, which are presented in Table 5. By far the most commonly observed keratopathy in both groups was corneal staining, which occurred more often in the central cornea than in the periphery, followed by redness and staining of the conjunctiva (20 cases in the medium-Dk group versus I0 cases in the high-Dk group over all visits). There were, 3 cases of corneal infiltrates and 13 cases of moderate- to-severe striae in the medium-Dk group, compared with the high-Dk group in which no infiltrative events and 2 cases of CLAK-grade striae occurred. Lens adherence that required changing lens parameters occurred in subjects .in each group. There were

Poke et a1 - Berkeley Contact Lens ~xtended Wear Study: Results

~ e d i u m - ~ k High-Dk

Clinician decision.lens-related Subject decision,lens-related Clinician decision.not lens-related Subject decision.not lens-related

3 month 6 month 9 month 12 month 3 month 6 month 9 month 12 month

Figure 1. After mndornization to the medium or high oxygen permeability (DK) study group, subjects could be dropped from the study by the clinician as 3 result of lens-related complications that could not be resolved, or for reasons unrelated to wearing the study lenses such as noncompliance with the studY protocol. Subjects themselves could elect to discontinue parriciparion for lens.relared discomfort if the clinician was unable to resolve the symprorn, or for various other reasons unrelated to lens wear (e.g., moving out of rhe area).

few or no cases of microcysts or polymegethisrn in either group. It is also interesting that in this study of RGPEW, we did not observe any cases of microbial keratitis, neovascularization, or acute red eye. which often occur in soft conract lens extended wear (SCLEW).'-'

Figure 3 shows the days from randomization to complete discontinuation of EW for the 21 subjects who were terminated from the study because of type 4 CLAK, along with the specific causes of failure. In total. 16 subjects in the medium-Dk group were terminated before 12 months of follow-up because of keratopathy or discomfort, whereas only high-Dk subjects failed for reasons directly related to lens wear. Of the 16 medium-Dk group terminations, 5 subjects failed for persistent lens adherence (i.e.. aher having successfully adapted to EW in stage 2 while wearing the high-Dk adaptation lenses). whereas 1 of the 5 failures in the high-Dk group was the result of adherence. All five cases of persistent lens adherence in the medium-Dk group eventually required complete discontinuation of EW, whereas four of the five adherence cases in the high-Dk group were resolved successfully. Although these clinical observations are suggestive of a greater risk of some lens-related complications. EW failure. or both in the medium-Dk group, censoring resulting from non-CL related factors makes survival analytic methods necessary for unbiased com- pikrisons of the two Dk groups.

Association of Hypoxic Dose with Ocular Complications

In this section. we present the survival analyses for time to initial onset of CLAK for each of the four definitions of CLAK. The number of elapsed days was calculated from the date of dispensing of the randomized study lenses to the date the clinician fint observed a CLAK event, which could have occurred at a regularly scheduled clinical assessment or an unscheduled visit; subjects who never exhibited such complications are right-censored at their last observation time 12 months or fewer from randomization. We analyzed the initial onset of CLAK for each subject rather than multiple CLAK events because of the difficulty in distinguishing unique events from recurrence of previously treated conditions and because attempts to ameliorate the first observed keratopathy often included disnlption of the EW schedule for varying periods of time. Furthermore, termination from the study because of a CLAK event censors such subjects with respect to possible subsequent events, but is not independent of the risk of such subsequent CLAK events as is required for the use of standard survival analytic techniques.

Figure 4 depicts the survival curves characterizing the p robf ' bility of remaining free from complications under the two different levels of hypoxia induced by the study lenses. The height of a

Ophthalmology Volume 108, Number 8, August 200 1

r Visit (mos.) Figure 2. Shown is the change from baseline in refractive sphere in the medium and high oxygen permeability (DK) groups at each visit. The ban: represent the least squares means and 95% confidence intewals. There was no significant difterence between DK groups in refractive sphere change from baseline (prefitring) measurements. There was no significant change across visits for either group. D = diuprers.

survival curve at any given point represents the estimated proba- months after the dispensing of the randomized study lenses and bility that a subject will remain free of complications until at least may have experienced CLAK early on that was not detected until that point. Under all four definitions of CLAK, the survival curves the first regularly scheduled quarterly follow-up visit: after that for the two Dk groups are virtually identical. The noticeable drop time, more opportunities may have been available to detect CLAK in the survival curves after 3 months in part may be the result of onset, because the clinician could note adverse symptoms at the our examination schedule: subjects were generally not seen for 3 3-month visit and request unscheduled visits or encourage the

Table 4. Cases of Contact Lens-associated Keraeopachy in the Two Lens Oxygen Permeability Groups

Medium Oxygen Permeability Group (n = 103) High Oxygen Permeability Group (n = 98)

Follow-up Visit (mos) 3 6 9 . I2 Total 3 6 9 I2 Tutcll

Number of active subjects 79 73 66 64 79 67 64 6 1 CLAK type I (slit-lamp I I 24 16 15 66 9 21) IS 9 56

hndings only) CLAK type 2 (including 1 I 45 I9 7 82 12 43 14 7 76

change in corneal curvature. refractive error) .

CLAK cype 3 (including 12 48 18 7 $5 13 47 13 5 -. rb subclinical lindings)

CLAK Gpe 4 (including 15 5 1 i6 7 S9 16 48 I2 5 S I s discomfort)

L+ CLAK = contact lens-associated keratopathy. Shown are the number of subjects presenting with CLAK in eirhtr eye for the hnt rime during their course ut' follow-up. CLAK discoveretl ilr :In r~nxheduled visit is included in the count for the nominal iullow-up visit clo~est ro date tri initial tletection.

Documents

The Berkeley Contact Lens Extended Wear Study: part I