Corneal Transplantation for Keratoconus

A Registry Study

Thu-Lan Kelly, PhD; Keryn A. Williams, PhD; Douglas J. Coster, DSc, FRANZCO; for the Australian Corneal Graft Registry

Author Affiliations: Department of Ophthalmology, Flinders University, Adelaide, Australia. Group Information: A list of the Australian Corneal Graft Registry contributors appears at /dspace/handle/2328/1723.

Objective: To determine factors influencing longterm graft survival and visual acuity in 4834 eyes of 4060 patients receiving their first penetrating corneal graft for keratoconus.

Methods: A large cohort study was performed from a national register of corneal grafts in which data were recorded prospectively and analyzed retrospectively. Main outcome measures were graft survival and Snellen visual acuity. Follow-up extended up to 23 years.

Results: Kaplan-Meier survival rates of first grafts for keratoconus were 89%, 49%, and 17% at 10, 20, and 23 years, respectively. After 15 years, the graft survival rate was no better than that of all other penetrating grafts (P=.36). Multivariate risk factors influencing failure of first grafts for keratoconus included time to suture removal, postgraft uveitis or microbial keratitis, corneal vascularization prior to or after graft, geographic location of surgery and follow-up (center effect), recipient age at graft, occurrence of rejection episodes, graft size, and surgeon workload. The timing of bilateral grafts made no difference to the risk of rejection. A Snellen visual acuity of 20/40 or better at the most recent follow-up was recorded in 74% of grafts.

Conclusions: Penetrating grafts performed for keratoconus exhibited better visual outcome and graft survival than grafts performed for other indications. However, the Kaplan-Meier survival rate of first penetrating grafts for keratoconus was 17% at 23 years after graft and had not plateaued at this time, indicating that young patients are likely to need 1 or more repeated grafts during their lifetime. Arch Ophthalmol. 2011;129(6):691-697. Published online February 14, 2011. doi:10.1001/archophthalmol.2011.7

KERATOCONUS IS A COMMON indication for penetrating keratoplasty.1,2 Generally, the survival of grafts performed for keratoconus is better than that of grafts performed for other indications.1-7 In recent years, lamellar corneal transplantation has been promoted as an alternative to penetrating keratoplasty for treating keratoconus.8 It is thus timely to review the outcome of penetrating corneal transplantation for this condition.
Excellent short-term9,10 and longterm11- 13 survival rates have been reported in single-center studies, but such studies may not reflect outcomes when the procedure is practiced in the broader community. Registries, in contrast, generate long-term outcome data across a wide range of clinical situations and are useful to fill evidence gaps14 in the absence of evidence from randomized clinical trials.
The Australian Corneal Graft Registry is a large multicenter registry that follows corneal grafts performed in various settings across Australia, reflecting varying surgeon preferences for case selection, surgical technique, and postoperative management in the real world. We examined graft survival and best-corrected visual acuity (BCVA) for 4834 patients recorded as undergoing their first penetrating keratoplasty for keratoconus from the registry’s inception in 1985 to November 2009.



Individual surgeons across Australia handle the consent process for each patient according to local legislative requirements, permitting information to be lodged with the Australian Corneal Graft Registry. The Institutional Ethics Committee of Flinders University oversees the operations of the registry, which are carried out in accordance with the Declaration of Helsinki.


All corneal grafts performed in Australia since May 1985 have been reported to the registry. Follow-up data are collected from 634 surgeons and other health care practitioners at 12-month intervals until graft failure or until the death or loss to follow- up of the patient. Missing data are sought directly from the surgeon or eye bank as appropriate. The information collected by the registry has been reported elsewhere.15 Patient death is tracked using a national database of deaths. Among eyes with followed penetrating grafts, loss to follow-up amounts to 30% and recipient deaths account for 24%.


Graft failure was defined as edema and loss of clarity in a previously thin, transparent graft or as irremediable astigmatism with or without recurrent keratoconus. Primary graft nonfunctions were defined as grafts that never cleared in the immediate postoperative period. Any eye that was regrafted was classified as having a previous graft failure, regardless of the reason. A rejection episode was defined as the development of inflammation, an anterior chamber reaction with corneal infiltrates (including subepithelial infiltrates) and spreading corneal edema (resulting from endothelial, epithelial, or stromal rejection), or a rejection line (epithelial or endothelial) in a previously thin, transparent graft. There were no instances of simultaneous bilateral rejection. Eyes with progressive edema in the absence of inflammation were classified as exhibiting corneal endothelial cell failure of unknown cause, not as rejection. Bestcorrected visual acuity was reported with the Snellen chart, with the recipient using visual aids (that is, spectacles or a contact lens) to achieve best correction, irrespective of whether the patient was tolerant of this correction in day-to-day living.


At the census date of November 2009, 19 529 penetrating grafts had been recorded, with 15 543 followed up on at least 1 occasion. Follow-up was available for 4834 first penetrating grafts in a given eye performed for keratoconus in 4060 patients, 774 (19%) of whom had bilateral penetrating grafts. The frequency of bilateral grafts changed with the length of time after graft: at 5 years or less after graft, patients were more likely to have unilateral grafts than bilateral grafts (P.001), whereas at more than 6 years after graft, there was no difference in the frequency of bilateral and unilateral grafts (P=.07). Follow- up ranged from 1 day to 23 years (median, 35 months). Compared with the remaining cohort of penetrating grafts, eyes with grafts for keratoconus had greater loss to follow-up (21% vs 54%, respectively) but fewer deaths reported (41% vs 3%, respectively). The median time between transplantation in the first and second eyes in the 774 patients with bilateral penetrating corneal grafts was 32months (range, 0 days to 20 years).


Statistical analysis was performed using Stata version 9 statistical software (StataCorp LP, College Station, Texas), with the significance level set at P.05. The Pearson 2 test was used to compare observed frequencies of graft failure, rejection, and vision characteristics in grafts for keratoconus, with expected frequencies generated using the proportions in the cohort of all penetrating grafts. The Mann-Whitney U test was used to compare age at graft and time to rejection. The relative risk for rejection in bilateral grafts was calculated using epidemiological 22 tables. Where appropriate, mean (standard deviation) or 95% confidence interval was reported. Graft survival among groups was compared with Kaplan-Meier plots, using the log-rank statistic to test statistical significance. Trial time was defined as the time from graft to failure for failed grafts and to the last follow-up for surviving grafts. Loss to follow- up was incorporated into Kaplan-Meier analysis. Variables that were significant in univariate survival analysis were included in multivariate analysis to calculate adjusted risk factors, controlled for by potential confounders. A Cox proportional hazards model clustered by patient was chosen. For adequate statistical power, only risk factors with 25 grafts or more in each category were included.16,17 confidence interval, 0.63-1.65; P=.95). Further, there was no difference in the unadjusted risk of rejection between the first and second grafted eyes (P=.19).

Figure 1. Kaplan-Meier graft survival plots for first penetrating grafts for keratoconus and for all other penetrating corneal grafts (numbers represent the number of grafts at risk).

Figure 2. Kaplan-Meier plots for first penetrating grafts for keratoconus with 1 or more episodes of rejection compared with those without any episode of rejection (numbers represent the number of grafts at risk).



Keratoconus was the most common indication for graft in the total cohort of penetrating keratoplasties (Table1). There was a significant difference (P.001) in age at graft for patients with first grafts for keratoconus (median, 30 years; range, 8-88 years) compared with the remaining cohort of penetrating grafts (median, 70 years; range, 0-99 years). First penetrating grafts for keratoconus exhibited significantly better survival (89% and 49% at 10 and 20 years after graft, respectively) than did penetrating grafts performed for other indications (P.001) (Figure 1).
Overall, 7% of first penetrating grafts for keratoconus failed over 23 years, in 19% of cases from irreversible immunological rejection (Table 2). By comparison, 23% of the total cohort of penetrating grafts failed, 29% of these from rejection. There was a significant difference in the frequency of failed grafts for keratoconus compared with all penetrating grafts (2 1=473.0; P.001) as well as in the frequency of rejected grafts (2 1=10.7; P=.001). Rejection (reversible or irreversible) in first penetrating grafts for keratoconus occurred most frequently in the first year after graft, with 48% of first rejection episodes occurring within 1 year and 90% within 4 years after graft. Time to rejection ranged from 10 days to 21 years (median, 1 year). Rejection was a significant risk factor in graft survival (P.001) (Figure 2). In patients with bilateral grafts, there was no difference in the risk of rejection after the second eye was grafted when the time between bilateral grafts was longer than 1 year compared with 1 year or less (relative risk=1.02; 95%


Figure 3. Kaplan-Meier graft survival plots for first penetrating grafts for keratoconus and for all other penetrating grafts that have survived at least 15 years from the time of the graft (numbers represent the number of grafts at risk).


A total of 235 first penetrating grafts for keratoconus (62%) in 215 patients survived 15 years or longer (median, 17 years). The median recipient age at surgery for these long-surviving grafts was 32 years (range, 11-77 years), similar to all grafts for keratoconus. The Kaplan- Meier survival rate was 17% (95% confidence interval, 2%-46%) at 23 years. For long-surviving grafts performed for any indication, there was no significant difference in survival between patients receiving grafts for keratoconus and the remaining cohort (P=.36) (Figure 3). The most common reasons for graft failure were unspecified cause, astigmatism, rejection, and recurrent keratoconus (Table 2).


Adjusted risk factors for graft failure were determined using a Cox proportional hazards model (Table 3). Variables included in the final model were age at graft, geographic location (center), surgeon experience (those performing 200 grafts), corneal vascularization at or after graft, graft size, refractive surgery after graft, inflammation in the past or at graft, follow-up arrangements, rejection episodes, uveitis, and microbial keratitis or stitch abscess after graft. Donor age was not found to influence graft survival.
To determine whether risk factors for graft survival changed over time, multivariate analysis was performed over the complete follow-up period and at 5-year intervals (Table 4). As the time after graft increased, the number of risk factors in each subset decreased. No risk factors for failure were identified for the 39 grafts surviving 20 years or longer. After 15 years, the major risk factor for graft survival or failure was the center effect: the worst center had a hazard more than 7 times that of the best center.


Figure 4. Visual outcome in first penetrating grafts for keratoconus. Best-corrected visual acuity (BCVA) was measured with spectacles and/or a contact lens preoperatively (A) and at the most recent follow-up (B). CF indicates counting fingers; HM, hand motion; LP, light perception; and NLP, no light perception.

Figure 5. Proportion of eyes with best-corrected visual acuity of 20/40 or better over time, measured with spectacles and/or a contact lens, at the most recent follow-up. Numbers of eyes at each time interval are shown above the data points. Error bars indicate standard error of the mean.


The proportion of eyes with a first penetrating graft for keratoconus that achieved a BCVA of 20/40 or better was 74% at the most recent follow-up compared with 8% preoperatively (Figure 4). The median preoperative BCVA was 20/200 (range, no light perception to 20/16), with contact lens intolerance being the reason for transplantation in eyes with good preoperative BCVA. The median postoperative BCVA at the last follow-up was 20/30 (range, no light perception to 20/13). The median change in BCVA after penetrating keratoplasty was an improvement of 6 lines (range, 10 lines worse to 13 lines better). Of eyes with a preoperative BCVA of 20/20 or better, 63% had worse BCVA, 26% had the same BCVA, and 11% had better BCVA at the most recent follow-up.
Overall, visual outcomes were significantly different in penetrating grafts performed for keratoconus compared with all other penetrating grafts (Table 5). Spectacles and/or contact lenses were prescribed for 61% of eyes grafted for keratoconus compared with 48% in all other penetrating grafts (2 1=130.1; P.001). Major astigmatism (5 diopters) wasmore prevalent in eyes grafted for keratoconus compared with all other penetrating grafts (2 1=39.0; P.001). At the last follow-up, significantly more eyes grafted for keratoconus compared with all other penetrating grafts had achieved BCVA of 20/40 or better (74% vs 45%, respectively; 2 1=782.6; P.001) and an improvement in BCVA postoperatively (2 2=269.4; P.001).
Because BCVA may stabilize over several years after graft, the proportion of patients with BCVA of 20/40 or better at the last follow-up was plotted for patients over time (Figure 5). Smaller numbers at increasing years after graft caused greater variability in the data. The proportion of eyes with BCVA of 20/40 or better at the last follow-up was 70% or greater until 11 years after graft, remaining greater than 59% up to 18 years after graft. For patients with grafts surviving longer than 15 years, BCVA of 20/40 or better at the last follow-up was achieved by 142 eyes (63%), while 82 eyes (37%) had BCVA worse than 20/40.



Within a large national multicenter registry, graft survival for eyes that received a first penetrating graft for keratoconus was significantly better than for the remaining cohort of penetrating grafts, and the proportion of grafted eyes with BCVA of 20/40 or better increased from 8% preoperatively to 74% at the most recent follow-up. Risk factors for failure of grafts for keratoconus over a follow-up period of 23 years were similar to those for all penetrating grafts.1 The number of risk factors for graft failure decreased with longer graft survival.
Consistent with many reports over the years,1-3,18 occurrence of rejection episodes in eyes grafted for keratoconus was a significant risk factor for graft failure. While most rejection episodes occurred soon after graft (90% in the first 4 postoperative years), some still occurred years later, with the longest time to first rejection being 21 years. Malbran and Ferna´ndez-Meijide19 found a higher incidence of rejection if a bilateral graft for keratoconus was performed within 1 year of the first graft. In contrast, we found that the timing of bilateral grafts made no difference to the risk of rejection after the second eye was grafted. We previously reported that patients with bilateral corneal grafts who have had a graft rejection episode in 1 eye were then at significantly greater risk for having a rejection episode in the other corneal transplant. 20 However, in the current cohort of low-risk patients with first grafts for bilateral keratoconus, there was no significant difference in the unadjusted risk of rejection between the first and second grafted eyes.
With respect to graft size, grafts of intermediate diameter appeared to have the best outcomes. Large grafts lie closer to the limbus with its load of recipient antigenpresenting cells than do intermediate grafts. The reason for the increased risk of failure of small-diameter grafts is unknown. Caseload also influenced graft survival significantly: surgeons who had registered more than 200 penetrating keratoplasties achieved significantly better graft survival. Of interest, advanced recipient age (60 years at graft) but not donor age was associated with significantly worse outcome in multivariate analysis. This effect was independent of the occurrence of rejection episodes and cannot have been due to the influence of regrafts in the same eye, given that the analysis was performed on first grafts in any one eye. Possibly some unidentified factor associated with the length of time older patients have had keratoconus may be responsible for poorer graft survival in these recipients.
While grafts for keratoconus showed better survival rates compared with survival of grafts for other indications for the first 15 postoperative years, after this time graft survival appeared independent of indication for graft. Penetrating grafts in high-risk eyes may be more likely to fail prior to 15 years, whereas all long-surviving grafts may be equally at risk for graft failure from late endothelial cell loss.21-24 In grafts performed for keratoconus, late graft failures were more often attributed to unspecified causes, recurrent keratoconus,11 or astigmatism25 and less often to rejection. For grafts surviving 15 years or longer, geographic location (center effect) was a risk factor for failure, with the worst center having more than 7 times the risk of graft failure than the best center. Efforts to tease out the basis of the center effect have thus far been unsuccessful. Donor- and eye bank–related factors do not influence graft survival significantly, and we have no evidence to suggest that patient populations with keratoconus differ across different geographic regions. All surgeons contributing data to the registry do so using an identical, paper-based form. Long-term survival for grafts with normal (rather than impaired) recipient endothelium was estimated by Borderie et al26 to be 41%±3% at 20 years, falling to 3%±1.0% at 30 years. Because most individuals with keratoconus would be expected to have normal recipient endothelium,23 the 20-year graft survival rate of 49% (95% confidence interval, 40%-57%) from our study compares well with the predictions of Borderie and colleagues.
A major weakness of registry studies is loss to followup, which is higher in patients with keratoconus than in the remaining cohort. There may be many patients with long-surviving grafts who have not attended a follow- up appointment for years. As patients with a failed graft are likely to seek medical attention, loss to follow- up may lead to an underestimate of graft survival at longer survival times. Further, as more data are collected by the registry, especially after 30 years following the graft when long-surviving grafts are more likely to fail, the data may become even more biased toward the reporting of failed grafts, so that the rate of failure is overestimated.
The principal purposes of corneal transplantation in patients with keratoconus are to improve visual function and reduce disability. In this study, 86% of grafts for keratoconus were performed to improve vision, of which BCVA is one measure. Visual acuity is difficult to standardize in community practice. Even when meticulously measured, BCVA may not be relevant to the daily functioning of the patient. A contact lens may be tolerated for the short time while BCVA is measured, but the patient, especially one with keratoconus, may not be prepared to use the lens with optimal refractive correction at other times. However, despite its limitations, BCVA is adequate to confirm the improvement in vision after penetrating keratoplasty for keratoconus. At the most recent follow-up, a BCVA of 20/40 or better was achieved by 74% compared with 8% preoperatively. However, for eyes with a preoperative BCVA of 20/20 or better, only 11% showed improved BCVA at the last follow- up, possibly because of regression to the mean.27 Patients with good preoperative BCVA, although less likely to achieve better BCVA postoperatively, may accept a reduction in BCVA as an alternative to wearing a contact lens. Overall, a higher proportion of eyes grafted for keratoconus exhibited a BCVA of 20/40 or better compared with all penetrating grafts, but this gain was achieved at the expense of a greater need for refractive aids and with significantly worse major astigmatism. Astigmatism is a particular problem with grafts for keratoconus26 and may be a cause of late graft failure. 28 In patients with keratoconus, visual acuity may take several years to stabilize after graft9; in this study, relatively stable visual acuity was not observed until about 5 years after transplantation.
Small single-center case series have reported longterm survival rates of grafts for keratoconus of 85% from 112 eyes11 and 93% from 125 eyes13 at 25 years. Registries contain a broader range of data than single-center studies and allow a more complete analysis of the factors that influence graft survival. However, the potential for underreporting graft survival because of loss to follow-up must be acknowledged. We report that in 4834 eyes grafted for keratoconus, Kaplan-Meier graft survival rates were 89%, 49%, and 17% at 10, 20, and 23 years, respectively. For long-surviving penetrating grafts such as for keratoconus, graft survival rates from our study and estimated by Borderie et al26 suggest that overall there is a low probability of graft survival after 30 years. Factors such as surgical skill and experience may increase a graft’s life span: a best-case Cox model incorporating the lowest hazard ratios for risk factors such as the center effect and surgeon caseload predicted survival rates of 85% and 46% at 20 and 23 years, respectively. Surgeons with the most experience should be favored when keratoplasty is unavoidable. A young person with a penetrating corneal graft is likely to require repeated keratoplasty in decades to come. Such repeated grafts tend to exhibit shorter survival than the first graft,1,2 and for young patients, especially those with excellent BCVA, alternatives to surgery are preferable.

Submitted for Publication: August 3, 2010; final revision received November 23, 2010; accepted December 8, 2010.
Published Online: February 14, 2011. doi:10.1001 /archophthalmol.2011.7
Correspondence: Keryn A. Williams, PhD, Department of Ophthalmology, Flinders Medical Centre, Bedford Park, South Australia 5042, Australia (keryn.williams@flinders
Author Contributions: Dr Williams had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analyses.
Financial Disclosure: None reported.
Funding/Support: This research was supported by the Australian Organ and Tissue Donation Authority. Dr Williams was supported by a fellowship from the Australian National Health and Medical Research Council.
Role of the Sponsors: The sponsoring institutions had no role in the design and conduct of the study; in the collection, management, analysis, or interpretation of the data; or in the preparation, review, or approval of the manuscript.
Additional Contributions: Marie Lowe,BSc, and Vicky Jones provided data entry and assistance with the analyses.


1. Williams KA, Lowe M, Bartlett C, Kelly TL, Coster DJ; All Contributors. Risk factors for human corneal graft failure within the Australian corneal graft registry. Transplantation. 2008;86(12):1720-1724.

2. Williams KA, Esterman AJ, Bartlett C, Holland H, Hornsby NB, Coster DJ. How effective is penetrating corneal transplantation? factors influencing long-term outcome in multivariate analysis. Transplantation. 2006;81(6):896-901.

3. Thompson RW Jr, Price MO, Bowers PJ, Price FW Jr. Long-term graft survival after penetrating keratoplasty. Ophthalmology. 2003;110(7):1396-1402.

4. Wagoner MD, Gonnah S, Al-Towerki AE; King Khaled Eye Specialist Hospital Cornea Transplant Study Group. Outcome of primary adult penetrating keratoplasty in a Saudi Arabian population. Cornea. 2009;28(8):882-890.

5. Tan DTH, Janardhanan P, Zhou H, et al. Penetrating keratoplasty in Asian eyes: the Singapore Corneal Transplant Study. Ophthalmology. 2008;115(6):975- 982, e1.

6. Claesson M, Armitage WJ. Ten-year follow-up of graft survival and visual outcome after penetrating keratoplasty in Sweden. Cornea. 2009;28(10):1124- 1129.

7. Dandona L, Naduvilath TJ, Janarthanan M, Ragu K, Rao GN. Survival analysis and visual outcome in a large series of corneal transplants in India. Br J Ophthalmol. 1997;81(9):726-731.

8. Watson SL, Ramsay A, Dart JK, Bunce C, Craig E. Comparison of deep lamellar keratoplasty and penetrating keratoplasty in patients with keratoconus. Ophthalmology. 2004;111(9):1676-1682.

9. Lim L, Pesudovs K, Coster DJ. Penetrating keratoplasty for keratoconus: visual outcome and success. Ophthalmology. 2000;107(6):1125-1131.

10. Javadi MA, Motlagh BF, Jafarinasab MR, et al. Outcomes of penetrating keratoplasty in keratoconus. Cornea. 2005;24(8):941-946.

11. Pramanik S, Musch DC, Sutphin JE, Farjo AA. Extended long-term outcomes of penetrating keratoplasty for keratoconus. Ophthalmology. 2006;113(9):1633- 1638.

12. Jensen LB, Hjortdal J, Ehlers N. Longterm follow-up of penetrating keratoplasty for keratoconus. Acta Ophthalmol. 2010;88(3):347-351.

13. Fukuoka S, Honda N, Ono K, Mimura T, Usui T, Amano S. Extended long-term results of penetrating keratoplasty for keratoconus. Cornea. 2010;29(5):528- 530.

14. Dreyer NA, Garner S. Registries for robust evidence. JAMA. 2009;302(7):790-791.

15. Williams KA, Lowe MT, Bartlett CM, Kelly L, Coster DJ, eds. The Australian Corneal Graft Registry 2007 Report. Adelaide, Australia: Flinders University Press; 2007.

16. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient, I: introduction and design. Br J Cancer. 1976;34(6):585-612.

17. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient, II: analysis and examples. Br J Cancer. 1977;35(1):1-39.

18. Ing JJ, Ing HH, Nelson LR, Hodge DO, Bourne WM. Ten-year postoperative results of penetrating keratoplasty. Ophthalmology. 1998;105(10):1855-1865.

19. Malbran ES, Ferna´ndez-Meijide RE. Bilateral vs unilateral penetrating graft in keratoconus. Ophthalmology. 1982;89(1):38-40.

20. Williams KA, Kelly TL, Lowe MT, Coster DJ; All Contributors to the Australian Corneal Graft Registry. The influence of rejection episodes in recipients of bilateral corneal grafts. Am J Transplant. 2010;10(4):921-930.

21. Armitage WJ, Dick AD, Bourne WM. Predicting endothelial cell loss and longterm corneal graft survival. Invest Ophthalmol Vis Sci. 2003;44(8):3326-3331.

22. Patel SV, Hodge DO, Bourne WM. Corneal endothelium and postoperative outcomes 15 years after penetrating keratoplasty. Am J Ophthalmol. 2005;139 (2):311-319.

23. Patel SV, Diehl NN, Hodge DO, Bourne WM. Donor risk factors for graft failure in a 20-year study of penetrating keratoplasty. Arch Ophthalmol. 2010;128 (4):418-425.

24. Zadok D, Schwarts S, Marcovich A, et al. Penetrating keratoplasty for keratoconus: long-term results. Cornea. 2005;24(8):959-961.

25. de Toledo JA, de la Paz MF, Barraquer RI, Barraquer J. Long-term progression of astigmatism after penetrating keratoplasty for keratoconus: evidence of late recurrence. Cornea. 2003;22(4):317-323.

26. Borderie VM, Boëlle P-Y, Touzeau O, Allouch C, Boutboul S, Laroche L. Predicted long-term outcome of corneal transplantation. Ophthalmology. 2009; 116(12):2354-2360.

27. Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. Int J Epidemiol. 2005;34(1):215-220.

28. Lim L, Pesudovs K, Goggin M, Coster DJ. Late onset post-keratoplasty astigmatism in patients with keratoconus. Br J Ophthalmol. 2004;88(3):371-376.