Laser in situ keratomileusis versus long-term contact lens wear: Decision analysis

Transcription

1 ARTICLE Laser in situ keratomileusis versus long-term contact lens wear: Decision analysis Hall T. McGee, MD, MS, William D. Mathers, MD PURPOSE: To compare the risk for vision loss from laser in situ keratomileusis (LASIK) versus long-term contact lens wear using decision analysis. SETTING: Oregon Health & Science University, Portland, Oregon, USA. METHODS: Based on previously published data, a decision tree was constructed using Markov modeling to calculate the probability of vision loss from LASIK versus contact lenses over time. An outcome variable, visual acuity adjusted life-year (VALY), was defined to account for timing and severity of vision loss. The VALYs were tallied over a 30-year simulation to determine whether the risk for vision loss with contacts exceeded that with LASIK. Sensitivity analyses were performed to test the effect of changes in key variables including probability of contact lens related keratitis, probability of ectasia after LASIK, and probability of early post-lasik vision loss. RESULTS: The model s conclusions were highly sensitive to changes in several variables tested, especially risk for post-lasik ectasia, risk for early vision loss after LASIK, and risk for contact lens related keratitis (which correlated with type of lenses worn). Rigid gas-permeable lenses were safer than LASIK in every analysis. The safety of LASIK exceeded that for daily-wear soft lenses only when assumptions were most favorable to LASIK, whereas the safety of LASIK always exceeded that for extended-wear lenses except when assumptions were least favorable to LASIK. CONCLUSION: This decision analysis found that the risk for vision loss associated with LASIK and with long-term contact lens wear might be closer than generally assumed. J Cataract Refract Surg 2009; 35: Q 2009 ASCRS and ESCRS Contact lenses and laser refractive surgery are popular alternatives to spectacles for correcting refractive errors. In 2005, there were 36 million contact lens wearers (3 million new wearers) in the United States, 1 and approximately 8 million people in the U.S. have Submitted: February 19, Final revision submitted: May 15, Accepted: May 19, From a private practice (McGee), Pittsburgh, Pennsylvania, and Casey Eye Institute (Mathers), Oregon Health & Science University, Portland, Oregon, USA. Neither author has a financial or proprietary interest in any material or method mentioned. Supported in part by an unrestricted grant to Casey Eye Institute from Research to Prevent Blindness, New York, New York, USA. Corresponding author: William D. Mathers, MD, Casey Eye Institute, 3375 Southwest Terwilliger Boulevard, Portland, Oregon , USA. [email protected]. had laser in situ keratomileusis (LASIK) (1.3 to 1.4 million new procedures per year). 2 Although there are risks associated with LASIK and with contact lenses, the classic assumption is that laser surgery is inherently riskier than contact lens wear. However, this assumption has never been formally tested. In 2005, Hammond et al. 3 reported the results of the United States Army excimer laser keratorefractive surgery program, documenting the incidence of complications and the visual outcomes in more than eyes. Several large epidemiologic studies have evaluated contact lens wearers and documented the risk for infectious keratitis and subsequent vision loss. 4 9 Comparing the results of these 2 alternatives to spectacle correction, Mathers et al. 10 commented in 2006 that the risk for vision loss with refractive surgery may be lower than that with a lifetime of contact lens use. Although previous studies evaluated the costeffectiveness of refractive surgery, 11,12 this is, to our knowledge, the first report of a formal decision analysis comparing the risk for losing corrected distance 1860 Q 2009 ASCRS and ESCRS Published by Elsevier Inc /09/$dsee front matter doi: /j.jcrs

2 LASIK VERSUS LONG-TERM CONTACT LENS WEAR 1861 visual acuity (CDVA) with LASIK and the risk with contact lens wear in patients with low to moderate myopia. MATERIALS AND METHODS Decision Tree The decision tree for this model was built (Figure 1) and analyzed with Markov modeling using the healthcare module of TreeAge Pro software (TreeAge Software, Inc.). 13 The decision tree begins with a patient with low to moderate myopia (defined as up to 6.00 diopters) who desires not to wear glasses. The first node represents the decision between LASIK and contact lenses. This then leads to 2 separate Markov models represented by their own subtrees, 1 for LASIK and 1 for contact lenses. Markov modeling, also referred to as state-transition modeling, is useful for modeling changes in health states over time. In this model, the LASIK subtree identifies 3 possible health states after surgery: (1) no visual acuity loss, (2) early postoperative visual acuity loss, and (3) subsequent development of ectasia. In this model, if patients have early postoperative loss of visual acuity or develop ectasia, they stay in that state forever with no possibility of returning to the no visual acuity loss state. Conversely, it remains possible for some patients initially assigned to the no visual acuity loss state to slip from the no visual acuity loss group into the ectasia group every year. The contact lens subtree functions similarly to the LASIK subtree. Patients are assigned to 1 of 4 potential health states: (1) no visual acuity loss, (2) moderate visual acuity loss, (3) severe visual acuity loss, and (4) very severe visual acuity loss. All patients are initially assigned to the category of no visual acuity loss. Every year (including during the first year), there is a small chance the patient will develop infectious keratitis. Cheng et al. 4 report that 13% of those with keratitis suffer vision loss. Those who had vision loss transition according to published probabilities are placed in moderate, severe, or very severe categories. Vision loss is again considered an absorbing state in this subtree, as in the LASIK subtree, with the implication being that patients would stop wearing their lenses and that their vision would get neither better nor worse. Each visual acuity state has an associated numerical reward that is applied at the end of the cycle. The model was allowed to iterate for 30 cycles. Literature Review The literature was searched using Ovid Medline (1950 to present) to identify English-language papers with data on the incidence and severity of vision loss in large populations of contact lens patients and LASIK patients. There are several such papers in the contact lens literature, which was recently reviewed by Keay et al. 5 Although there are many studies of the risk for vision loss after LASIK, most are smaller than the contact lens studies and there is no clear consensus on the actual risk. Thus, in the current study, the risk for vision loss from LASIK was defined based on the results of the largest refractive study identified. 3 A sensitivity analysis was then performed to account for the disparate reports of the risk for vision loss. This LASIK paper 3 is from the U.S. military s refractive surgery program, which (unlike the contact lens studies) evaluated a distinct population in a closed system and, as such, is large enough to capture data on many thousands of refractive surgery patients. A study by Cheng et al. 4 reported visual acuity outcomes after contact lens associated keratitis in a way that was amenable to numerical analysis; thus, those estimates were used. Overall, the findings of Cheng et al. are similar to those in other studies of the subject. 5 9 Figure 1. Decision tree showing the decision between LASIK and contact lenses as 2 separate Markov models. The possible visual states are the first branches to the right of the LASIK or contact lens Markov nodes (circled letter M). Subsequent branches show the possible transition paths to the next visual states, which can be the same as the first or different (LASIKZlaser in situ keratomileusis; pectasiazprobability of ectasia; pkeratitis_clzprobability of keratitis with contact lens wear; pmodvaloss_clzprobability of moderate visual acuity loss with contact lens wear; psev- VALoss_CLZprobability of severe visual acuity loss with contact lens wear; LASIK; pva_loss_lasikzprobability of visual acuity loss after LASIK; VAZvisual acuity).

3 1862 LASIK VERSUS LONG-TERM CONTACT LENS WEAR Table 1. Visual acuity states after contact lens associated keratitis and LASIK by the published definitions and by how the outcomes were defined for this analysis in Snellen and logmar visual acuity notation. Visual State Published Definition Definition in Present Dtudy (Snellen VA) LogMAR Equivalent VA* Contact lenses Moderate VA loss 20/30 20/60 20/ Severe VA loss 20/70 or worse 20/ Very severe VA loss Light perception 20/ LASIK Vision loss More than 1 line of VA lost (no patient worse than 20/40) 20/ Ectasia Pre-management 20/108 Post-management 20/37 20/ LASIKZlaser in situ keratomileusis; VAZvisual acuity *The rewards used in this study were the negative of the logmar-equivalent visual acuity listed here Visual Acuity Adjusted Life-Years Appropriate accounting for vision loss requires consideration of when the vision loss events occur as well as the severity of the vision loss. Some levels of vision loss are worse than others, and delaying vision loss until later in life is preferable to having it occur earlier. Therefore, this analysis models risk in terms of years of life with vision loss and adjusts for the severity of the vision loss. For the model, the visual acuity adjusted life-years (VALY) unit was devised. The VALY unit is analogous to a qualityadjusted life-year, which uses utility values associated with a health state multiplied by the number of life years spent in that health state. 14 In the present study, each year of life spent with vision loss was weighted according to the relative severity of the vision loss by multiplying by the negative of the respective logmar visual acuity value (Table 1). Adjustments were made for severity of vision loss. For LASIK, 2 main ways to lose visual acuity were modeled: (1) early postoperative vision loss and (2) development of ectasia. For contact lenses, all vision loss was assumed to be the result of infectious keratitis, which could be moderate, severe, or very severe. For early postoperative LASIK vision loss, the published definition was more than 1 line of vision loss (worse than 20/25) but no patient had worse than 20/40 CDVA after LASIK in the Hammond et al. 3 study. Therefore, an average of 20/30 (2 lines of vision loss) was used for LASIK-associated vision loss. Because Hammond et al. report a very low risk for vision loss from LASIK (probability of losing more than 1 line of CDVA was or 0.06%), this variable was analyzed in a sensitivity analysis over a large range to reflect different opinions on the matter. There are many estimates of the risk for post-lasik ectasia, although most are not specific to low to moderate myopia, in which it is generally thought to be rare. One literature review estimated 1 case per 5000 patients (R.D. Stulting, MD, Ectasia: How Serious Is the Problem? presented at the annual meeting of the American Academy of Ophthalmology, Las Vegas, Nevada, November 2006). In this study, this estimate was used as the baseline with a range of 0.0% to 0.2% (1 in 500). Next, that effect was spread over 30 years, with a brisk reduction in risk over time (40% per year) so the bulk of the effect accrued in the first 5 years, as is generally observed clinically. The severity of vision loss from post-lasik ectasia was estimated to be comparable to severe vision loss from contact lens related keratitis (20/70 CDVA) based on a report of CDVA, with ectasia being 20/108 before ectasia management and 20/37 afterward (89% of cases treated with rigid gas-permeable [RGP] contact lenses, spectacles, or no correction). 15 For contact lenses, Cheng et al. 4 defined moderate vision loss as 20/30 to 20/60 (20/30 used in present study) and severe vision loss as 20/70 or worse (20/70 used in present study). In the severe vision loss category, Cheng et al. report 1 patient who was left with light perception only. This is a measure of stimulus perception, not visual acuity, and as such cannot be numerically analyzed on the same scale. Therefore, a category of very severe vision loss was included to account for this rare event, with the visual acuity analyzed being equal to 20/800. The probability of infectious (or presumed infectious) contact lens associated keratitis varies depending on the type of lens used (Table 2). 4,7 9 Even in cases of keratitis, most patients have no significant vision loss; however, approximately 13% do lose visual acuity. 4,9 The Snellen visual acuity measurements were converted mathematically to logmar values for the analysis. 16 The number of years with vision loss was then multiplied by the negative of the appropriate logmar weight (Table 1) to account for the relative severity of the vision loss. This yielded the main outcome measure, logmar VALY. Adjustment for Timing of Vision Loss An important component in comparing the risk for vision loss is when the vision loss might occur. Markov modeling naturally accounts for the timing of an event as points accumulate for each year spent with vision loss. In addition, this analysis covered a long time period (30 years), meaning that discounting was necessary to appropriately weight adverse events that occurred early versus those that occurred later. Discounted rewards were calculated according to the principle of net present value. In this case, an annual rate of 5% was used. A high discount rate would tend to favor contact lenses because vision loss from LASIK occurs earlier and, as a result, was discounted less. Therefore, sensitivity analysis was performed over the range of 0% to 7% as recommended elsewhere 17 to determine whether variation in the discount rate would materially affect the outcomes of the analysis. A half-cycle correction was used to account for the fact that vision loss events occur throughout the year, not just at the

4 LASIK VERSUS LONG-TERM CONTACT LENS WEAR 1863 Table 2. List of variables used to build the model. Range* (%) Variable Description Default Value (%) Low High Probability of early VA loss from LASIK (per event) Probability of ectasia from LASIK Probability of visually significant contact lens related keratitis (per year) Probability of any VA loss from keratitis 5 13 NA NA Probability of moderate VA Loss Probability of severe VA Loss Probability of very severe VA Loss Discount rate LASIKZlaser in situ keratomileusis; NAZnot applicable; VAZvisual acuity *Range of values tested in sensitivity analysis Range that corresponds to the values for rigid gas-permeable lenses at the low end and extended-wear lenses at the high end beginning or the end. Half-cycle correction applies half of a reward as an initial reward and the other half as a final reward, with a full reward applied incrementally at every intermediate stage. (Note: Half-cycle correction was used for all rewards except early visual acuity loss from LASIK, which was assumed to occur immediately postoperatively and was tabulated with a full initial reward and full incremental rewards.) Sensitivity Analysis This model was based on previously published reports, which differed substantially in their findings for LASIK and contact lenses. There are similarly heterogeneous practice patterns around the world, which may also lead to different outcomes. One way to account for this uncertainty is to perform sensitivity analysis to determine how sensitive the outcome of the model is to changes in 1 or more variable. Sensitivity analyses show the range of conclusions one might draw given different values for the variables as well as which variables have the greatest impact on the model. Table 2 shows the variables tested and the range over which they were analyzed. Also, some authors suggest that contact lens related keratitis is more common in novice wearers and declines in incidence as patients gain more experience. 5,18 Therefore, this was tested with a constant value for the risk for keratitis and with a variable risk for keratitis that decreased over time by 10% per year. RESULTS In this study, the decision between LASIK and contact lenses depended heavily on the set of parameters tested. For contact lenses, the mechanism of vision loss is via infectious keratitis and this correlates closely with the type of lens used. For LASIK, vision loss can occur for many reasons in the early postoperative period, or from post-lasik ectasia as a late complication. Wearing RGP lenses minimized the risk for vision loss, measured by VALYs lost. In no scenario tested was the risk for vision loss with RGP lenses higher than that with LASIK over the 30 years of simulation. Daily-wear soft contact lens wear had a more moderate risk for keratitis and vision loss, and that risk was higher with extended-wear lenses. The decision in those cases, therefore, depended more heavily on the other variables in the model. Figure 2 shows the results of 1-way sensitivity analyses of time. Low-risk RGP lenses never accumulated more risk than LASIK (Figure 2, top row). The risk with daily-wear soft contact lenses exceeded that with LASIK after approximately 15 years if the risk for ectasia was low (1 case per patients) (Figure 2, middle row, left) but never exceeded that with LASIK if the ectasia risk was higher (1 case per 1000 patients) (Figure 2, middle row, right). The risk with extendedwear soft contact lenses exceeded that with LASIK almost immediately when the risk for ectasia was low (Figure 2, bottom row, left) and also with a higher risk for ectasia (Figure 2, bottom row, right). If the true risk for ectasia were 1 case per 500 (0.20%), even extended-wear soft contact lenses would have lower risk for vision loss than LASIK (Figure 3). If the true risk for ectasia were low, however, the risk for vision loss with LASIK would be exceeded by extended-wear soft contact lenses, even with an early postoperative risk of 0.48% (almost 1 case in 200). On the other hand, the risk with dailywear soft contact lenses would exceed that with LASIK only under conditions most favorable to LASIK (Figure 3, J and M). The conclusions of this study were also sensitive to the rate used to discount the vision loss effects that occur over time. In this model, a high discount rate favored contact lenses, which tended to be associated with later-onset vision loss. Although this study used 5.0% as the default discount rate, the model shows that LASIK would be preferred over dailywear soft contact lenses if the discount rate were less than approximately 4.2% (Figure 4).

5 1864 LASIK VERSUS LONG-TERM CONTACT LENS WEAR Figure 2. One-way sensitivity analyses on time (years) showing cumulative number of VALYs per 1000 patients (DSCLsZdailywear soft contact lenses; EWSCLs Zextended-wear soft contact lenses; ptszpatients; RGPsZrigid gas-permeable contact lenses; VALYsZvisual acuity adjusted life-years). Low risk of ectasia depicted in the left column of graphs (0.01%) and high risk in the right column of graphs (0.1%). Sensitivity analysis of parameters related to contact lens associated vision loss showed that if contact lens-related keratitis were concentrated in novice wearers, the vision loss from those events would be incurred earlier as well. That tended to hurt the case for contacts. On the other hand, the severity of vision loss from keratitis was directly proportional to the VALYs lost from contact lens wear. The model was only relatively insensitive to changes in the variables tested in the sensitivity analysis (Table 2). The model was very sensitive to the probability of keratitis; however, the decision between LASIK and contact lenses was quantitatively affected by the timing and severity of vision loss. There were no alterations to the conclusions regarding whether contact lenses or LASIK would be preferred in this analysis. DISCUSSION To our knowledge, this is the first formal decision analysis comparing the safety of long-term contact lens wear with that of LASIK. Several major factors necessitate a decision analysis. First, there may be more events of vision loss with long-term soft contact lens wear than with LASIK. 10 Second, the events associated with LASIK occur earlier; thus, patients accumulate more time with vision loss. Third, the vision loss from contact lens associated keratitis can be more severe than vision loss with LASIK; therefore, the years of relatively worse disability must be weighted more heavily. These factors counteract each other, and intuition alone is insufficient to account for the variability. Indeed, contrary to intuition, this decision analysis shows that in some commonly occurring situations, contact lenses may be riskier than LASIK. Another finding was that RGP contact lenses are safe and never surpassed the risk associated with LASIK under any condition tested. Daily-wear soft contact lenses may be riskier than LASIK, but only under conditions most favorable to LASIK (ie, the probabilities of post-lasik vision loss and ectasia are low). Extended-wear soft contact lenses are known to be associated with a higher risk for visually

6 LASIK VERSUS LONG-TERM CONTACT LENS WEAR 1865 Figure 3. Four-way sensitivity analysis. The following 4 variables are shown: (1) time (x-axis of each panel); (2) probability of contact lens associated keratitis (y-axis of each panel) and the range, which encompasses the risk associated with RGP lenses on the low end through daily-wear soft contact lenses and up to extended-wear soft contact lenses on the high end (reference lines on each panel and arrows on right margin); (3) probability of VA loss from LASIK (columns), and (4) cumulative probability of post-lasik ectasia (rows). The blue areas show the parameters in which LASIK would have less loss of VALYs. The green areas show scenarios in which contact lenses would be preferred (CLZcontact lens; DSCLsZdaily-wear soft contact lenses; EWSCLsZextended-wear soft contact lenses; RGPsZrigid gas-permeable contact lenses).

7 1866 LASIK VERSUS LONG-TERM CONTACT LENS WEAR Figure 4. One-way sensitivity analysis of the discount rate (ptszpatients; VALYsZvisual acuity adjusted life-years). significant keratitis than RGP lenses or daily-wear soft lenses. In our model, extended-wear soft contact lenses had the highest risk of all lenses and accumulated more vision loss than LASIK under all conditions except those most favorable to contact lenses (ie, the probabilities of post-lasik vision loss and ectasia are high). The conclusions of this model were very sensitive to changes in some parameters but relatively insensitive to others. In particular, the model was very sensitive to changes in the probability of contact lens related keratitis, the probability of post-lasik ectasia, and the probability of early post-lasik vision loss. Fluctuations in the severity of vision loss after contact lens related keratitis had relatively little impact on the model. Similarly, it made relatively little difference whether the rate of contact lens related keratitis was constant or was made to decline with time. The current study offers a way of considering this issue; however, some facets to this decision were not addressed. For instance, patients may be more psychologically willing to accept vision loss related to contact lenses than vision loss after LASIK because of a sense of their own responsibility for the outcome. Also, a major advantage of contact lenses is the ability to change refractive correction easily. Although LASIK can often be repeated if necessary, the risks are such that it would not be recommended for small changes in refraction. On the hand, LASIK can provide better correction than contact lenses in certain circumstances, such as in patients with astigmatism who have unstable rotation of toric soft contact lenses. The likelihood of long-term spectacle independence is also an important consideration, although it is difficult to quantify and to compare in this sort of analysis. Also, although spectacle independence is generally the goal of refractive surgery, typical contact lens patients will rely on spectacles occasionally. Therefore, rates of spectacle independence were not included in the present model. This study analyzed visual acuity outcomes only. Visual acuity is often considered to represent quantity of vision as opposed to quality of vision. Both LASIK and contact lenses have been associated with glare, halos, star bursts, dry eye, and eye irritation. These side effects are difficult to quantify for the purposes of analysis. One way to address this issue would be with the concept of utility. Utility measurements would allow the calculation of quality-adjusted life-years for contact lens wearers and LASIK patients and pave the way for a true head-to-head cost-utility analysis, which would probably be the most definitive and externally comparable sort of analysis. 14 In conclusion, both contact lenses and LASIK can be safe and effective for correcting low to moderate myopia in the right patient. This decision analysis is the first to formally compare the risk for losing visual acuity with the 2 methods. From this analysis, which is based on previously published epidemiologic reports, it appears there may be some commonly occurring circumstances in which long-term contact lens wear could be riskier than LASIK surgery for low to moderate myopia. More study of this question is required to determine whether the confluence of those circumstances truly exists in practice. Nevertheless, it is a significant departure from conventional wisdom to think that the safety of LASIK could even approach that of contact lenses. REFERENCES 1. Barr JT. Contact lenses Contact Lens Spectrum 2006; 21(1): 26, 28, 30, Available at com/article.aspx?articlez Accessed July 21, Andrews M. A lighter, defter touch; years of refinement have made laser eye surgery better than ever. U.S. News and World Report, March 5, 2007; 59 61; posted online February 25, Available at: articles/070225/5laser.htm. Accessed July 15, Hammond MD, Madigan WP Jr, Bower KS. Refractive surgery in the United States Army, Ophthalmology 2005; 112: Cheng KH, Leung SL, Hoekman HW, Beekhuis WH, Mulder PGH, Geerards AJM, Kijlstra A. Incidence of contactlens-associated microbial keratitis and its related morbidity. Lancet 1999; 354: Keay L, Stapleton F, Schein O. Epidemiology of contact lensrelated inflammation and microbial keratitis: a 20-year perspective. Eye Contact Lens 2007; 33: Nilsson SEG, Montan PG. The annualized incidence of contact lens induced keratitis in Sweden and its relation to lens type and wear schedule: results of a 3-month prospective study. CLAO J 1994; 20: Poggio EC, Glynn RJ, Schein OD, Seddon JM, Shannon MJ, Scardino VA, Kenyon KR. The incidence of ulcerative keratitis

8 LASIK VERSUS LONG-TERM CONTACT LENS WEAR 1867 among users of daily-wear and extended-wear soft contact lenses. N Engl J Med 1989; 321: Schein OD, Glynn RJ, Poggio EC, Seddon JM, Kenyon KR. The relative risk of ulcerative keratitis among users of dailywear and extended-wear soft contact lenses. A case-control study; Microbial Keratitis Study Group. N Engl J Med 1989; 321: Schein OD, McNally JJ, Katz JL, Chalmers RL, Tielsch JM, Alfonso E, Bullimore M, O Day D, Shovlin L. The incidence of microbial keratitis among wearers of a 30-day silicone hydrogel extended-wear contact lens. Ophthalmology 2005; 112: Mathers WD, Fraunfelder FW, Rich LF. Risk of LASIK surgery vs contact lenses [letter]. Arch Ophthalmol 2006; 124: Javitt JC, Chiang Y-P. The socioeconomic aspects of laser refractive surgery. Arch Ophthalmol 1994; 112: Lamparter J, Dick HB, Krummenauer F. Clinical benefit, complication patterns and cost effectiveness of laser in situ keratomileusis (LASIK) in moderate myopia; results of independent meta analyses on clinical outcome and postoperative complication profiles. Eur J Med Res 2005; 10: Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Med Decis Making 1993; 13: Brown MM, Brown GC, Sharma S, Landy J. Health care economic analyses and value-based medicine. Surv Ophthalmol 2003; 48: Woodward MA, Randleman JB, Russell B, Lynn MJ, Ward MA, Stulting RD. Visual rehabilitation and outcomes for ectasia after corneal refractive surgery. J Cataract Refract Surg 2008; 34: Holladay JT. Visual acuity measurements [guest editorial]. J Cataract Refract Surg 2004; 30: Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-Effectiveness in Health and Medicine. New York, NY, Oxford University Press, 1996; Schein OD, Katz J. Comparing contact lens and refractive surgery risks [letter]. Arch Ophthalmol 2007; 125: ; reply by WD Mathers, First author: Hall T. McGee, MD, MS Private Practice, Pittsburgh, Pennsylvania, USA