Patient outcomes of refractive surgery

articles Patient outcomes of refractive surgery The Refractive Status and Vision Profile Oliver D. Schein, MD, MPH, Susan Vitale, PhD, MHS, Sandra D. Cassard, ScD, Earl P. Steinberg, MD, MPP ABSTRACT Purpose: To evaluate the performance of a questionnaire, the Refractive Status and Vision Profile (RSVP), in the assessment of patient outcomes following refractive surgery. Setting: Patients recruited from 5 refractive surgery practices. Methods: The RSVP was self-administered by patients before and 2 to 6 months after bilateral refractive surgery. Information on uncorrected visual acuity (UCVA), refractive error, and self-reported satisfaction with vision was also collected. Changes in total RSVP scores and in the scores of RSVP subscales (concern, functioning, driving, symptoms, optical problems, glare, and trouble with corrective lenses) were assessed. The relationship between change in the RSVP and subscale scores was assessed in relation to change in traditional clinical measures. The responsiveness of the RSVP to clinically meaningful changes in patients vision was measured by calculating its effect size. Results: One hundred seventy-six patients completed baseline and postoperative RSVPs and had bilateral refractive surgery. Postoperatively, 92.0% of patients had a UCVA of 20/40 or better in at least 1 eye. Fifteen percent had some worsening in the total RSVP score, and there was substantial variation in the proportion of patients who had worsening in particular subscale scores, ranging from 7.0% who reported worsening in trouble with corrective lenses to 41.5% who reported worsening in driving. Change in satisfaction with vision following surgery was correlated with change in the overall RSVP and subscale scores but not with change in refractive error. A significant worsening in 3 or more RSVP subscales was independently associated with an almost 6-fold (odds ratio 5.84, 95% confidence interval: 1.88,18.13) likelihood of patient report of dissatisfaction with vision, after adjusting for age, sex, preoperative refractive error, and postoperative UCVA. Low scores (ie, minimal dysfunction) on 2 of the RSVP subscales (physical functioning and optical problems) at baseline were predictive of poor postoperative patient outcomes. The RSVP was very sensitive to the intervention of refractive surgery (effect size of 1.2 to 1.4). Conclusions: The RSVP was able to detect clinically relevant changes in functional status and quality of life after refractive surgery. Change in the RSVP score was correlated with change in patient report of satisfaction and was predictive of postoperative patient satisfaction. The RSVP provides a valuable new metric to assess outcomes of refractive surgery. J Cataract Refract Surg 2001; 27:665 673 2001 ASCRS and ESCRS 2001 ASCRS and ESCRS 0886-3350/01/$ see front matter Published by Elsevier Science Inc. PII S0886-3350(01)00844-6

There has been a rapid evolution in refractive surgery techniques and the size of the population having refractive procedures. This phenomenon has emphasized the need for reliable methods to assess the outcomes of refractive surgery. Although the success or failure of refractive surgery has usually been assessed by uncorrected vision and refractive accuracy, the importance of incorporating the patient s perspective into the evaluation of outcomes of medical and surgical treatments is now generally recognized. The patient s perspective is commonly assessed by questionnaires that examine functional status, quality of life, and symptoms in a standardized way. Several questionnaires have been rigorously validated in patients with cataract. Two such questionnaires are the Activities of Daily Vision 1,2 and the VF-14. 3 5 Both questionnaires, which focus on functional status related to vision, have been shown to be sensitive in the detection of clinically meaningful changes following cataract surgery and to provide information predictive of the outcome of cataract surgery that is as powerful as that provided by the traditional predictors of age and ocular comorbidity. 2,4,5 These questionnaires, however, are unlikely to be well suited to assess patient function related to refractive error and its correction because they focus on activities that depend on central or peripheral vision. An individual with a corrected refractive error would likely achieve similar scores on such questionnaires, as would individuals without refractive errors who have normal vision. Most important, these questionnaires are insensitive to the limitations (visual, symptoms, quality of life, social) reported by patients who are considering refractive surgery. Accepted for publication March 8, 2001. From the Dana Center for Preventive Ophthalmology, Wilmer Ophthalmological Institute (Schein, Vitale), the Departments of Medicine and Health Policy and Management (Steinberg), and a consulting practice (Cassard), Baltimore, Maryland, USA. Supported in part by NRSA EY07127: Clinical Trials Training Program (Vitale), NIH grant K24 EY00395 (Schein), and the Burton E. Grossman Program for Preventive Ophthalmology (Schein). None of the authors has a financial interest in any product mentioned. Reprint requests to Oliver D. Schein, MD, 116 Wilmer Building, Johns Hopkins Hospital, 600 North Wolfe Street, Baltimore, Maryland 21287-9019, USA. E-mail: oschein@jhmi.edu. OUTCOMES OF REFRACTIVE SURGERY We developed a questionnaire, the Refractive Status and Vision Profile (RSVP), to measure functional status and quality of life in individuals with refractive errors. The cross-sectional evaluation of this questionnaire 6 demonstrates that the RSVP and its composite subscales are valid as measured by standard psychometric techniques. This cross-sectional assessment, however, did not examine whether the RSVP is capable of detecting clinically significant change in functional status, quality of life, or satisfaction with vision after refractive surgery. To address this issue, a prospective evaluation of the RSVP was undertaken. Patients and Methods The cross-sectional validation of the 42-item RSVP 6 was performed in 550 patients from 5 refractive surgery practices and 1 optometric practice recruited between May and December 1997. Patients enrolled in the validation study were eligible for the prospective assessment if they completed a baseline RSVP and subsequently had refractive surgery (either photorefractive keratectomy [PRK] or laser in situ keratomileusis [LASIK]) in both eyes at the participating refractive surgery practices (see Appendix). A copy of the RSVP questionnaire was sent to eligible patients to complete between 2 and 6 months after surgery in the second eye. If no response was received, the questionnaire was sent again. If there was still no response, up to 5 attempts were made by telephone to contact the patient to complete the questionnaire. In addition to completing the RSVP, patients were asked to rate their satisfaction with vision on a 5-point scale (very dissatisfied, dissatisfied, neither satisfied nor dissatisfied, satisfied, or very satisfied) and to rate their concern about their vision and overall health. Clinical and demographic data were obtained by the study coordinator at each center. The study was approved by the human studies committees at the participating sites, and each patient gave oral consent to participate. Statistical Analysis Change in the overall RSVP and the subscale scores was first determined by calculating the difference between the preoperative and postoperative scores. Using 666 J CATARACT REFRACT SURG VOL 27, MAY 2001

this method, a patient would be classified as improved or worsened if there was any change in his or her scores. This approach would misclassify patients whose change in score was due to noise or imprecision of the RSVP itself. Therefore, a more conservative approach was also used in which a statistically significant change in a score was defined as a change in score of at least twice the standard error of measurement as obtained from the baseline test retest assessment. 6 One of the RSVP subscales relates to expectations. The 2 questions in this subscale address patients anticipated postoperative tolerance of less-than-perfect vision. Since it does not make sense to include such questions postoperatively, this subscale was dropped from the presurgical versus postsurgical comparisons. A Cronbach was computed for the 40-item RSVP and was found to be 0.92 (ie, there was no change in the internal consistency of the questionnaire after removing this subscale). Several bivariate and multivariate analyses were performed to examine change in the total RSVP score (S) and its subscale scores in relation to other variables. The association between preoperative refractive error and mean changes in the RSVP and its subscale scores were analyzed using an analysis of variance with a test for linear trend. The association between change in satisfaction with vision and change in the RSVP (S) and its subscale scores was assessed using the Spearman rank correlation coefficient. Multiple logistic regression was used to predict postoperative dissatisfaction with vision using changes in RSVP subscale scores and postoperative uncorrected visual acuity (UCVA) or need for corrective lenses as predictors. A combined outcome was created (postoperative dissatisfaction with vision or significant worsening of 3 or more RSVP subscale scores), and bivariate associations were examined in relation to patient characteristics. Multiple logistic regression was used to assess the association between patients preoperative characteristics and the combined outcome variable to better understand the predictors of a poor patientreported outcome following refractive surgery. All analyses were performed using SAS, version 6.12 (SAS Institute). The sensitivity of the RSVP in detecting clinically meaningful changes following surgery (responsiveness) was assessed using 2 techniques. The standard measure of responsiveness is the effect size, calculated by incorporating data from all patients and dividing the mean change in the RSVP (S) and its subscale scores by the standard deviation of the scores at baseline. 7 A second technique for calculating the effect size is the responsiveness statistic of Guyatt. 8 This statistic also uses the mean change in the measure as the numerator, but its denominator is the standard deviation of the change in the score of patients who are unchanged over time based on clinical criteria. This technique acknowledges that some patients may have improvement or worsening in a score even though they are clinically unchanged. To assess the effect size of the RSVP by this technique, the stable group was defined by the test retest performance of the 16 myopic patients assessed for reproducibility in the original cross-sectional validation of the RSVP. 6 Although there is no absolute scale for determining whether a questionnaire is responsive, according to Cohen, 9 it is reasonable to regard an effect size of 0.2 as small, of 0.5 as medium, and of 0.8 or higher as large. Results Of the 346 patients from the 5 centers who were eligible for the study, 176 (51%) completed a postoperative RSVP. The baseline characteristics of those completing versus those not completing the RSVP were compared. There were no significant between-group differences in preoperative refractive error, satisfaction with vision, or total RSVP score. Patients completing the postoperative RSVP were more likely to be women (64% versus 52%, P.02) and slightly older (mean age 39 years versus 36 years, P.004). Tables 1A and 1B show the baseline clinical characteristics of the study population and the standard clinical outcomes of refractive error and UCVA, respectively. Hyperopic patients were 3% of the total. There was a significant proportion of patients with moderate to high myopia (34% with 6.0 to 9.9 diopters [D] and 12% with 10.0 D or higher). Forty-seven patients had PRK and 53, LASIK. Postoperatively, 92% of patients had a distance acuity of 20/40 or better in their better eye and 76% had this acuity in their worse eye. Seventy-two percent of worse eyes were within 1.0 D of emmetropia and 91% were within 2.0 D. Thirty-three percent of patients indicated they were dissatisfied or very dissatisfied with their vision preoperatively, and 15% indicated this postoperatively. Sixty- J CATARACT REFRACT SURG VOL 27, MAY 2001 667

Table 1A. Baseline characteristics (n 176). Characteristic Percentage Age (years) 18 39 51 40 49 37 50 12 Sex Female 64 Corrective lenses Contact lenses only 18 Glasses and contact lenses 40 Glasses only 42 Refractive error (D) Hyperopic 3 0to 2.9 10 3.0 to 5.9 41 6.0 to 9.9 34 10.0 or higher 12 BCVA 20/20 87 D diopter; BCVA best corrected visual acuity Table 1B. Postoperative clinical outcomes. Postoperative Outcome Worse Eye (%) Better Eye (%) Refractive error (SE, D) (n 152) 0.5 44.7 73.7 0.6 to 1.0 27.0 13.8 1.0 to 2.0 19.1 11.2 2.0 9.2 1.3 UCVA (n 151) 20/20 28.5 58.3 20/25 20/40 47.7 33.8 20/50 20/80 17.2 6.6 20/100 20/160 3.3 0.0 20/200 3.3 1.3 SE standard error; D diopter; UCVA uncorrected visual acuity eight percent of those who were dissatisfied or very dissatisfied preoperatively indicated they were satisfied or very satisfied postoperatively. The changes in the total RSVP score (S) and in each individual subscale score are shown in Table 2. Eightyfive percent of patients had some improvement in the total RSVP score. Using the more conservative statistical definition of change, the total RSVP score improved in 66.5% of patients, worsened in 4.5%, and did not change in the remaining 29.0%. Substantial variation was observed in both the proportional and absolute change in individual subscale scores of the RSVP. The largest proportional change was observed in trouble with corrective lenses; the mean fell from 35.0% preoperatively to 7.4% postoperatively (on a scale of 0 to 100). Large proportional improvements in mean scores were also observed for concern (45.0% and 26.7%, respectively), functioning (15.9% and 7.4%, respectively), and symptoms (19.6% and 10.8%, respectively). Large variation was seen in the proportion of those who improved significantly (range 27.4% to 74.4%) or worsened significantly (range 2.3% to 29.5%). Despite a good outcome reflected by Snellen acuity, significant worsening was reported in the subscale concern by 7.4% of patients, functioning by 5.9%, driving by 29.5%, symptoms by 12.7%, optical problems by 19.9%, and glare by 16.3%; 2.3% reported more trouble with corrective lenses. Approximately 52.0% of patients had no significant worsening in any subscale, 26.0% had worsening in 1, 22.0% had worsening in 2 or more, and 14.7% had significant worsening in 3 or more. Table 3 shows correlations between postoperative change in satisfaction with vision, change in refractive error, and change in the RSVP and its subscale scores. While there was no correlation between change in refractive error in either eye and change in satisfaction, significant correlations with change in satisfaction were observed for the RSVP (S) and most of its subscales. There was no relationship between either preoperative refractive error or change in refractive error and change in the RSVP and its subscales. Table 4 evaluates the responsiveness of the RSVP (S) and its subscales using 2 external measurements of improvement, satisfaction with vision and not needing spectacle or contact lens correction postoperatively. While the effect size of the overall RSVP (S) was large, the responsiveness of the individual subscales varied substantially: driving was consistently the least responsive due to approximately equal numbers of patients improving and worsening, and glare and optical problems were only modestly responsive. Concern and trouble with corrective lenses were consistently the most responsive 668 J CATARACT REFRACT SURG VOL 27, MAY 2001

Table 2. The RSVP subscales and overall S: preoperative, postoperative, and change statistics. RSVP Subscale Preoperative Postoperative Mean SD Range Mean SD Range % Any Improvement % Any Worsening % Significant Improvement % Significant Worsening Concern 45.0 18.1 4.2 87.5 26.7 16.8 0 75 78.9 14.3 62.9 7.4 Functioning 15.9 14.2 0 65 7.4 14.5 0 86.4 70.4 19.5 40.2 5.9 Driving 24.3 22.7 0 100 23.7 21.7 0 100 38.0 41.5 28.3 29.5 Symptoms 19.6 16.6 0 75 10.8 10.7 0 75 59.0 27.1 44.6 12.7 Optical problems 11.8 13.6 0 62.5 10.3 13.5 0 87.5 42.8 34.9 27.7 19.9 Glare 22.4 17.3 0 75 19.0 16.1 0 75 45.2 33.7 28.9 16.3 Trouble w/corrective lenses RSVP score* Total S (40 items) 35.0 17.3 7.1 81.3 7.4 16.3 0 75 86.5 7.0 73.7 2.3 24.0 10.6 3.2 54.5 13.2 10.5 0 59.7 84.7 15.3 66.5 4.5 *RSVP and all subscale scores rank from 0 to 100; lower scores indicate less dysfunction. Change greater than 2 standard error of measurement subscales. The responsiveness of the RSVP and its subscales was also evaluated, stratified by preoperative refractive error. With the exception of physical functioning, in which there was a trend toward greater responsiveness for greater degrees of preoperative refractive error, the responsiveness of the subscales did not vary by the amount of preoperative refractive error. Figure 1 is a Venn diagram illustrating the relationship among 3 measures of less than optimal refractive surgery outcome: the traditional clinical outcome of UCVA worse than 20/40, dissatisfaction with vision, and a proposed new measure, a significant worsening in 3 or more of the RSVP subscales. As is evident from the diagram, these 3 measures assess different dimensions of patient outcome. Five patients fell into the intersection of the 3 measures. The strongest predictors of dissatisfaction after refractive surgery are felt to be poor distance UCVA and an ongoing need for spectacles or contact lenses. Since these variables are highly correlated, 3 separate analyses were done (Table 5), each adjusting for age and sex; the first included UCVA, the second, the use of corrective lenses for distance; and the third, the use of lenses for distance or near as independent variables. In each model, significant worsening of 3 or more subscales of the RSVP was independently associated with dissatisfaction with vision following refractive surgery. In each model, worsening in 3 or more subscales was associated with an approximately 6-fold Table 3. Spearman correlations between change in satisfaction with vision and changes in different outcome measures (n).* Change in Outcome Measure Change in Satisfaction with Vision Change in refractive error better eye 0.10; P.2375 (n 150) Change in refractive error worse eye 0.06; P.4443 (n 150) Change in RSVP (S) 0.59; P.0001 (n 174) Change in concern 0.44; P.0001 (n 173) Change in physical functioning 0.48; P.0001 (n 167) Change in driving 0.37; P.0001 (n 164) Change in symptoms 0.32; P.0001 (n 164) Change in optical problems 0.33; P.0001 (n 164) Change in glare 0.33; P.0001 (n 164) Change in trouble w/corrective lenses 0.25; P.0011 (n 169) *Denominator varies because the small proportion of missing responses varies by question. J CATARACT REFRACT SURG VOL 27, MAY 2001 669

Table 4A. Responsiveness of RSVP and its subscales using improved postoperative satisfaction with vision as the measure of outcome. RSVP Scale Effect Size* (n 91) Guyatt s Responsiveness Statistic* (n 91) Concern 25.6/17.9 1.43 25.6/6.7 3.82 Physical functioning 13.5/13.9 0.97 13.5/7.2 1.88 Driving 7.4/21.7 0.34 7.4/8.3 0.89 Symptoms 14.0/18.0 0.78 14.0/5.8 2.41 Optical problems 6.5/15.6 0.42 6.5/4.4 1.48 Glare 9.4/18.7 0.50 9.4/7.2 1.31 Trouble with corrective lenses 33.0/19.1 1.73 33.0/8.8 3.75 Total RSVP score (S) 16.4/11.1 1.48 16.4/3.9 4.21 *Numerator is mean change in subscale score (or Total S) among those who reported any improvement in satisfaction with vision after surgery. Denominator is standard deviation of preoperative subscale score (or Total S) among those who reported any improvement in satisfaction with vision after surgery. Denominator is standard deviation of change in subscale score (or Total S) among stable group of myopes who did not have surgery. 6 Table 4B. Responsiveness of RSVP and its subscales using postoperative need for corrective lenses as the measure of outcome. RSVP Scale Effect Size (n 153) Guyatt s Responsiveness Statistics (n 153) Concern 20.2/18.1 1.12 20.2/6.7 3.01 Physical functioning 8.8/13.8 0.64 8.8/7.2 1.22 Driving 0.7/21.2 0.03 0.7/8.3 0.08 Symptoms 8.9/16.4 0.54 8.9/5.8 1.53 Optical problems 1.8/13.7 0.13 1.8/4.4 0.41 Glare 4.1/17.0 0.24 4.1/7.2 0.57 Trouble with lenses 31.1/17.6 1.77 31.1/8.8 3.53 Total RSVP score* (S) 12.1/10.4 1.16 12.1/3.9 3.10 *Numerator is mean change in subscale score (or Total S) among those who reported any improvement in satisfaction with vision after surgery. Numerator is mean change in subscale score (or totals) among those who reported no use of distance lenses (glasses or contacts) after surgery. Denominator is standard deviation of preoperative subscale score (or Total S) among those who reported no use of distance lenses (glasses or contacts) after surgery. Denominator is standard deviation of change in subscale score (or Total S) among stable group of myopes who did not have surgery. 6 excess risk of reporting dissatisfaction with vision postoperatively. To test the hypothesis that responses to a preoperataive RSVP might be helpful in predicting postoperative patient-reported outcomes, a final analysis was performed. A poor patient outcome was defined as a postoperative report of being dissatisfied with vision or having a significant reduction in 3 or more of the RSVP subscales. In these analyses, each RSVP subscale and the total RSVP (S) were examined independently as predictors. Table 6 summarizes the results of these analyses. Age, sex, and preoperative refractive error were not predictive, although age greater than 45 years was consistently associated with an odds ratio of approximately 1.5 (ie, those over 45 years were 50% more likely to be dissatisfied). Three of the preoperative RSVP subscales (expectations, physical functioning, and symptoms) were associated with approximately twice the likelihood of a poor postoperative outcome. In these multivariate analyses, physical functioning (P.06) and symptoms (P.05) were the most predictive. Specifically, patients who reported good physical functioning and few symptoms (lowest decile for each preoperatively) were more likely to be dissatisfied or report a significant worsening in 3 or more RSVP subscales postoperatively. 670 J CATARACT REFRACT SURG VOL 27, MAY 2001

Table 5. Predictors of postoperative dissatisfaction with vision. Model A (n 151) Model B (n 174) Model C (n 174) Predictor Odds Ratio (95% CI*) Odds Ratio (95% CI) Odds Ratio (95% CI) Age 45 years 1.75 (0.59, 5.16) 1.52 (0.53, 4.37) 0.93 (0.30, 2.92) Female 2.20 (0.69, 6.98) 1.89 (0.61, 5.86) 2.06 (0.72, 5.94) Preoperative refractive error 1.02 (0.87, 1.19) 1.03 (0.87, 1.20) 1.00 (0.86, 1.16) 3 RSVP subscales w/significant worsening 5.84 (1.88, 18.13) 6.96 (2.21,21.87) 6.20 (2.10, 18.26) Model A Postoperative UCVA worse than 20/40 in either eye 3.54 (1.23, 10.19) Model B Wear distance lenses 11.68 (3.72, 36.73) Model C Wear distance or reading lenses 4.86 (1.60, 14.80) *CI confidence interval Spherical equivalent, eye with less refractive error Change greater than 2 standard error of the mean variability 0.01 P.05 P.01 Table 6. Association of preoperative RSVP subscales or total S with postoperative poor outcome* (dissatisfaction with vision or significant worsening on 3 or more subscales) (n 174). Subscale Odds Ratio (95% CI ) Concern 0.79 (0.25, 2.53) Expectations 2.03 (0.84, 4.88) Physical functioning 2.35 (0.96, 5.76) Driving 1.06 (0.44, 2.56) Symptoms 2.41 (1.00, 5.79) Optical problems 1.41 (0.64, 3.10) Glare 1.17 (0.44, 3.09) Trouble with lenses 0.32 (0.07, 1.47) Total S 1.02 (0.31, 3.37) *Adjusted for age, sex, and baseline refractive error CI confidence interval Each subscale or Total S was dichotomized at the first decile with the lowest decile 1 and the remaining values 0 P.06 P.05 Discussion The most common indices for measuring outcomes following refractive surgery are UCVA and residual refractive error. Patient-reported outcomes have either not been assessed or have been limited to reports of satisfaction with vision. The RSVP and its subscales have been shown to exhibit several properties that make it a useful outcome measure that would complement the traditional clinical assessments. First, the overall RSVP and its subscales have been shown to demonstrate good psychometric performance in a large, cross-sectional validation. 6 Second, these measures are highly responsive to refractive surgery. Third, changes in the RSVP and its subscales are correlated with postoperative change in patient-reported satisfaction with vision. Change in refractive error was not correlated with this outcome. Fourth, as illustrated in the Venn diagram (Figure 1), the use of Snellen acuity, self-reported satisfaction with vision, and the RSVP as outcome measures do not identify the same group of patients as successes of the surgical intervention. The RSVP clearly provides additional information about patient outcomes that is not captured by simply assessing patient satisfaction or UCVA. Moreover, as shown in the multivariate analyses, significant worsening in 3 or more RSVP subscales was independently associated with dissatisfaction with vision. This association is of the same magnitude as that for postoperative UCVA and need for corrective lenses. J CATARACT REFRACT SURG VOL 27, MAY 2001 671

Figure 1. (Schein) Venn diagram illustrating minimal overlap among 3 different measures of outcomes following refractive surgery. The magnitude of improvement in scores on the RSVP and its subscales did not vary in relation to the magnitude of the initial preoperative refractive error. Although this finding appears counterintuitive, it is analogous to the findings reported for cataract surgical patients when the VF-14 was evaluated as one of multiple outcome measures following cataract surgery. 3 For cataract surgery, it was found that the magnitude of perceived functional improvement was not greater in those with worse visual acuity preoperatively (eg, 20/200 versus 20/50). This indicates that the problems with visual function experienced by those who opted for cataract surgery were not reflected by Snellen acuity but by perceived functional need. One might infer, therefore, that the magnitude of refractive error is not adequate as a sole measure to characterize the potential change in patients functional status or perceived limitations (eg, symptoms, difficulties with glasses/contact lenses) that might be anticipated following refractive surgery. Change in the overall RSVP score was more closely associated with change in satisfaction with vision than were change in refractive error or change in the RSVP subscale scores. Yet, use of the total RSVP score alone is not optimal. For example, it is possible that the total score might show improvement even if there were a significant (ie, important to the patient) decrement in 1 or more of the individual subscales. Therefore, it would be preferable to use and report the individual subscales in any characterization of patients having refractive surgery. It would be potentially very informative to learn, for example, that an intervention might provide a large benefit in functional status but a worsening in symptoms or glare. Retention of the subscales allows this possibility. The overall RSVP and a number of its subscales were found to be very responsive to refractive surgery as assessed by 2 measures of effect size. The finding that a particular subscale has a small effect size does not necessarily imply that the subscale provides no useful information in the assessment of the intervention. For example, the driving subscale consistently was shown to have the smallest effect size. Yet, as shown in Table 2, an equal and large proportion of patients reported improvement and worsening of driving ability following refractive surgery. The effect size uses as its numerator the overall average change in a group. Thus, if equal numbers have large worsenings and large improvements, the 2 extremes will cancel, making the average (and the effect size) close to zero. The effect size is not a good measure of responsiveness when both improvement and worsening are equally possible. Clearly, one would not want to remove issues related to driving from an assessment of the outcomes of refractive surgery, even though this subscale has a low effect size. From the perspective of surgeons using the standard clinical outcomes of distance UCVA and achieved refractive error, the outcomes of this cohort of refractive surgery patients was very good, especially given the large proportion of patients with moderate to high myopia preoperatively. Yet, as demonstrated, 13% to 30% of patients experienced worsening in the RSVP subscales related to driving, symptoms, optical problems, and glare. From the patient perspective, therefore, there remains significant room for improvement in refractive surgery. The RSVP offers a strategy to improve our understanding of patient outcomes of refractive surgery. Specifically, it may offer insight into the sources of patient dissatisfaction that may be experienced postoperatively. Other standard measures such as UCVA, refractive error, or simply asking patients to rate their satisfaction with vision will not be helpful in this regard. The RSVP may also be helpful in the conduct of clinical trials related to the safety and effectiveness of refractive surgery techniques or devices. In this context, the RSVP would provide a useful mechanism to compare populations to ensure balance between groups at baseline in a randomized trial or to assess comparability of patient populations across studies. Improvement in vision-related quality of life or worsening in symptoms or function (eg, 672 J CATARACT REFRACT SURG VOL 27, MAY 2001

glare, trouble driving) are effectiveness and safety outcomes in their own right about which patients and their surgeons would like to be informed. The findings in this article should not be interpreted as an indication of patient outcomes of refractive surgery in general. The research was performed to evaluate the RSVP rather than to describe surgical outcomes by technique or other factors. The surgery performed in this cohort was done in 1997 at 5 centers, it was limited to those having surgery in both eyes, and the follow-up was incomplete. It is possible that one might find a greater benefit (eg, due to fewer patient-perceived limitations) from refractive surgery performed in the year 2000 or in a different group of patients. In addition to providing a mechanism to understand and quantify a spectrum of patient outcomes following refractive surgery, it is possible that the RSVP may help in the optimal selection of patients for refractive surgery. Our preliminary data suggest that patients with relatively little preoperative dysfunction in the domains of physical functioning and symptoms are more likely to be dissatisfied with vision or report significant worsening in 3 or more RSVP subscale scores postoperatively. These findings indicate that it may be possible to identify patients preoperatively with the RSVP who may be at a greater or lesser risk of operative success, as perceived by the patient. Data on more recent and larger cohorts will be necessary to confirm the findings reported here and to allow further development and refinement of a predictive model. Ultimately, it would be most useful if one were able to associate specific threshold RSVP or subscale scores preoperatively with the likelihood of different patient-reported outcomes postoperatively. In the interim, we conclude that the incorporation of the patient s perspective on health-related quality of life associated with a refractive error and its correction will provide a valuable new metric to assess patient outcomes of refractive surgery. In the future, such a metric will help provide a benchmark that will be useful in further improving these outcomes. References 1. Mangione CM, Phillips RS, Seddon JM, et al. Development of the Activities of Daily Vision Scale ; a measure of visual functional status. Med Care 1992; 30:1111 1126 2. Mangione CM, Orav EJ, Lawrence MG, et al. Prediction of visual function after cataract surgery; a prospectively validated model. Arch Ophthalmol 1995; 113:1305 1311 3. Steinberg EP, Tielsch JM, Schein OD, et al. The VF-14; an index of functional impairment in patients with cataracts. Arch Ophthalmol 1994;112:630 638 4. Steinberg EP, Tielsch JM, Schein OD, et al. National study of cataract surgery outcomes; variation in 4-month postoperative outcomes as reflected in multiple outcome measures. Ophthalmology 1994; 101:113 141; discussion by DM O Day, 1140 1141 5. Schein OD, Steinberg EP, Cassard SD, et al. Predictors of outcome in patients who underwent cataract surgery. Ophthalmology 1995; 102:817 823 6. Vitale S, Schein OD, Meinert CL, Steinberg EP. The Refractive Status and Vision Profile; a questionnaire to measure vision-related quality of life in persons with refractive error. Ophthalmology 2000; 107:1529 1539 7. Kazis LE, Anderson JJ, Meenan RF. Effect sizes for interpreting changes in health status. Med Care 1989; 27(suppl):S178 S189 8. Guyatt G, Walter S, Norman G. Measuring change over time: assessing the usefulness of evaluative instruments. J Chronic Dis 1987; 40:171 178 9. Cohen J. Statistical Power Analysis for the Behavioral Sciences, 2nd ed. Hillsdale, NJ, Lawrence Erlbaum Associates, 1988 Appendix Participating Refractive Surgery Practices Massachusetts Eye and Ear Infirmary, Boston, Massachusetts, USA: Dimitri Azar, MD, investigator, David Rees, study coordinator; Gimbel Eye Centre and the Gimbel Foundation, Calgary, Alberta, Canada: Howard Gimbel, MD, investigator; Maria Ferensowicz, study coordinator; Cornea & Laser Vision Institute, Teaneck, New Jersey, USA: Peter Hersh, MD, investigator; Bethann Hibbert, study coordinator; Lisa Schwartz, study coordinator; Washington Eye Physicians and Surgeons, Chevy Chase, Maryland, USA: Roy Rubinfeld, MD, center director; Kate Kelly, study coordinator; Wilmer Eye Institute, Baltimore, Maryland, USA: Terrence O Brien, MD, center director; Nada Jabbur, MD, investigator; Richard Schoen, OD, investigator; Rebecca Scarborough, study coordinator. J CATARACT REFRACT SURG VOL 27, MAY 2001 673