Economic impact of the advent of posterior lamellar ... - Science Direct

0 downloads 0 Views 191KB Size Report
Objective: To assess the cost-utility of posterior lamellar keratoplasty (PLK) techniques, including ... activity in North America.4,5 In past decades, improvements.
Economic impact of the advent of posterior lamellar keratoplasty in Montreal, Quebec Catherine Beauchemin,* MSc; Isabelle Brunette,{{ MD, FRCSP; He´le`ne Boisjoly,{{ MD, MPH; Ellen E. Freeman,{ PhD; Mihaela Popescu,{ MD; Jean Lachaine,* PhD ABSTRACT N RE´SUME´ Objective: To assess the cost-utility of posterior lamellar keratoplasty (PLK) techniques, including deep lamellar endothelial keratoplasty, Descemet stripping endothelial keratoplasty, and Descemet stripping automated endothelial keratoplasty, in the treatment of corneal endothelial diseases. Design: Cost-utility analysis based on a Markov model. Participants: Cohort of 100 patients waiting for corneal graft. Methods: This cost-utility analysis was performed from a Canadian health system perspective over a lifetime period. A Markov model was constructed to compare the cost per quality-adjusted life-year (QALY) associated with penetrating keratoplasty (PK) and PLK techniques. The model included all major health states relevant to patients scheduled for corneal transplant: waiting for transplant, surviving graft with or without complications, irreversible failure, noneligibility, and death. Transition probabilities among health states were obtained from published clinical trials. Costs considered were those associated with surgery, patients’ follow-up, and postsurgical complications. Number of QALYs was estimated presurgery and postsurgery using the Brown and Sharma conversion chart. Results: PLK proved to be more effective, providing more QALYs (+13 QALYs/100 patients), and it was less costly (2$68 792/100 patients) compared with PK. Deterministic and probabilistic sensitivity analyses confirmed the robustness of the base-case results. Conclusion: From a clinical and an economic standpoint, PLK in the management of patients waiting for corneal graft represents a preferred strategy compared with PK only. Objet : E´valuation de la rentabilite´ des techniques de ke´ratoplastie lamellaire poste´rieure (KLP), y compris la ke´ratoplastie lamellaire endothe´liale profonde, la ke´ratoplastie endothe´liale avec stripping de Descemet et la ke´ratoplastie endothe´liale automatise´e avec stripping de Descemet, pour le traitement des maladies endothe´liales de la corne´e. Nature : Analyse de rentabilite´ selon un mode`le de Markov. Participants : Une cohorte de 100 patients en attente d’une greffe de la corne´e. Me´thodes : Cette analyse de rentabilite´ s’inspire d’une perspective du re´gime de sante´ canadien pendant la dure´e de la vie. On a e´labore´ un mode`le de Markov pour comparer le couˆt par anne´e de vie ponde´re´e par la qualite´ (QALY) associe´ a` la ke´ratoplastie pe´ne´trante (KP) et aux techniques de KLP. Le mode`le comprend tous les principaux e´tats de sante´ pertinents aux patients devant recevoir une transplantation corne´enne : patients en attente, ayant surve´cu a` une greffe avec ou sans complications, e´chec irre´versible, non e´ligibles et de´ce´de´s. Les probabilite´s de transition entre les e´tats de sante´ ont e´te´ tire´es des publications sur les essais cliniques. Les couˆts conside´re´s ont e´te´ associe´s a` ceux de la chirurgie, du suivi des patients et des complications postchirurgicales. Le nombre de QALY a e´te´ estime´ avant et apre`s la chirurgie selon le tableau de conversion de Brown et Sharma. Re´sultats : La KLP s’est ave´re´e plus efficace, donnant plus de QALY (+13 QALY/100 patients), et moins couˆteuse (268 792 $ / 100 patients) comparativement a` la KP. Les analyses de sensibilite´, de´terministes et probabilistes, ont confirme´ la solidite´ des hypothe`ses de base. Conclusion : Du point de vue clinique et e´conomique, la KLP offre une strate´gie pre´fe´rable a` la KP seule pour traiter les patients en attente de greffe de la corne´e.

T

he cornea is the most commonly transplanted tissue, with more than 120 000 corneal transplants performed worldwide each year.1–3 Estimated at about 50 000 in the United States and 2500 in Canada, the annual number of corneal grafts shows the high corneal transplant activity in North America.4,5 In past decades, improvements

in surgical procedures have made penetrating keratoplasty (PK) one of the most successful transplants.6 Although PK is associated with high success rates, this type of transplant is limited by the shortage of corneal tissue and consequent lengthy wait times from diagnosis to surgery. In Canada, the wait time for corneal graft

From *the Faculty of Pharmacy; {the Department of Ophthalmology, University of Montreal, Montreal, Que.; and {the Department of Ophthalmology, Maisonneuve-Rosemont Hospital, Montreal, Que.

Correspondence to Catherine Beauchemin, MSc, Faculty of Pharmacy, University of Montreal, P.O. Box 6128, Station Centre-Ville, Montreal QC H3C 3J7; [email protected]

Presented at the annual meeting of the Canadian Ophthalmological Society in Toronto, Ont., June 2009

This article has been peer-reviewed. Cet article a e´te´ e´value´ par les pairs.

Originally received June 4, 2009. Final revision Dec. 9, 2009 Accepted Jan. 25, 2010 Published online May 11, 2010

Can J Ophthalmol 2010;45:243–51 doi:10.3129/i10-026

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

243

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. is a major challenge; the average wait time for corneal transplant was estimated at 51(SD 32) weeks in 2004.7,8 Wait status may be associated with anxiety, poor visual acuity, dissatisfaction with vision, ocular pain, and discomfort, and it wields a substantial impact on the patient’s quality of life.9 Posterior lamellar keratoplasty (PLK), which uses only the posterior portion of the donor cornea, represents a promising alternative in the treatment of endothelial corneal disease.10–13 PLK techniques, including deep lamellar endothelial keratoplasty (DLEK), Descemet stripping endothelial keratoplasty (DSEK), and Descemet stripping automated endothelial keratoplasty (DSAEK), offer significant advantages over standard PK for patients with endothelial disorders.13–22 First, the absence of corneal sutures leads to less astigmatism and fewer suture-related complications such as neovascularisation, inflammation, and infectious keratitis.13 Second, PLK leads to more rapid visual recovery than PK.18,23,24 Moreover, PLK may provide better accessibility to corneal graft because it permits the use of donor tissues that are unsuitable for PK because of anterior corneal disease or prior surgery.25,26 However, the benefits of PLK are limited to endothelial corneal conditions because this technique is unsuitable for the treatment of any nonendothelial conditions, such as bacterial scarring, keratoconus (KC), herpes simplex virus keratitis, and chemical burns. Thus, considering that patients with nonendothelial conditions would still receive a PK if PLK was implemented, the use of PLK in the management of patients with a corneal disorder may offer significant advantages, but only for those suffering from endothelial corneal disease. Although clinical results concerning PLK have been reported extensively in recent years, little is known about its economic impact. The purpose of this study was to assess the cost-utility of PLK in the management of patients waiting for corneal graft in Montreal, Que. METHODS Model design A Markov model was constructed to assess the costs per quality-adjusted life-year (QALY) associated with PLK compared with only PK. Because PLK is a relatively new and evolving procedure, all PLK techniques (DLEK, DSEK, and DSAEK) were combined, with the intention of collecting as much clinical and economic data as possible. The cost-utility analysis was performed over a lifetime horizon from a Canadian health-care system perspective. A hypothetic cohort of 100 patients waiting for a corneal graft was created according to clinical characteristics observed in medical practice. Based on a previous analysis of patients undergoing PK in Quebec, the proportion of women included in the cohort was slightly higher than the proportion of men (54.4% vs 45.6%), and the mean age of patients was 62 years.27 To better reflect the dynamic

244

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

surrounding the wait time for corneal graft in medical practice, patients’ preoperative diagnoses included, but were not limited to, pseudophakic bullous keratopathy, regraft, KC, and Fuchs’ endothelial dystrophy, which are leading indications for corneal transplant.28 Because KC patients are not candidates for PLK, they were not included in the PLK group. In fact, a hypothesis of the model assumed that all KC patients would have received PK even if PLK was implemented. The proportion of patients with KC was established according to the proportion of KC patients included in the corneal graft wait list at Maisonneuve-Rosemont Hospital (Que.) from 2000 to 2007 (n 5 486). The Markov model comprised all major health states relevant to patients scheduled for corneal transplant: waiting for transplant, surviving graft with or without complications, irreversible graft failure, noneligibility, and death (Fig. 1). The length of each Markov cycle was 1 year and the model continued to run until all patients reached the absorbing state, defined as death. All patients started in the waiting-for-transplant state and could move to other health states thereafter. Specifically, patients could stay in the waiting-for-transplant state if they did not receive a corneal graft at the end of a cycle. They could also become noneligible for the graft and stay in the noneligibility state until they reached the absorbing state. Moreover, patients who had received the expected graft could end up in either the surviving-graft-with-or-without-complications or the irreversible-graft-failure state. Postoperative complications such as graft rejection, glaucoma, microbial keratitis, endophthalmitis, and disc dislocation were considered in the first year in which patients entered the surviving-graftwith-or-without-complications state. Finally, patients could have experienced irreversible graft failure in the first postoperative year (transition from waiting for transplant to irreversible graft failure) or in subsequent years (transition from surviving graft with or without complications to irreversible graft failure). Transition probabilities Transition probabilities were estimated using various data sources. First, transition probabilities related to corneal graft

Fig. 1—Structure of the Markov model.

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. accessibility were determined by an analysis of Quebec Eye Bank data. Specifically, the probability that a patient remained in the waiting-for-transplant state, or conversely received a corneal graft, was determined by an analysis of 486 patients placed on the corneal transplant wait list at Maisonneuve-Rosemont Hospital from 2000 to 2007. Similarly, the probability of becoming noneligible for the corneal graft was determined by the same analysis. Reasons for noneligibility for the graft included patient’s refusal, progress of the disease, incidence of another general or ocular health condition, and waiting period longer than 6 years. In addition, transition probabilities related to postoperative graft survival and complications were obtained from the results of published clinical trials. The probabilities of experiencing irreversible graft failure or other postoperative complications over time have been presented elsewhere.29 Transition from a health state of the model to the absorbing state (death) was extracted from Canadian complete life tables for 2000–2002, according to the age and sex of the model population.30 All transition probabilities are presented in Table 1 and rates of postoperative complications are presented in Table 2. Costs Costs included in the model comprised those associated with surgery, patient follow-up, and postsurgical complications (Table 3). Specifically, the costs associated with corneal transplant included those related to ophthalmologist and anaesthesiologist fees and maintenance medications, and those incurred in hospital. Ophthalmologist and anaesthesiologist fees were obtained from the 2008 Re´gie de l’assurance maladie du Que´bec (RAMQ) manual for specialist physicians, whereas hospital costs were obtained from the accounting system of the Centre hospitalier de l’Universite´ de Montre´al. Costs incurred at the hospital were obtained

for PK, and they were considered to be the same for PLK. Costs related to maintenance medications were estimated from the list of reimbursed medications by the prescription drug plan (RAMQ). Medication use was determined by expert opinion (Isabelle Brunette and He´le`ne Boisjoly), according to the use of pharmacological agents in current medical practice. Costs associated with physician fees, hospitalizations, and maintenance medication did not differ among keratoplasty techniques. However, costs related to the Moria surgical instruments were added in the PLK group and were obtained from the supplier of the device (I-MED Pharma Inc, Dollard-des-Ormeaux, Que.). Costs related to these instruments were amortized according to the total number of procedures performed by one such instrument. In the base-case analysis, the model assumed that one Moria surgical instrument would perform a total of 500 procedures, which was the forecasted use over a 10-year period at Maisonneuve-Rosemont Hospital. Health-care expenditures associated with patient follow-up were obtained from the 2008 RAMQ manual for specialist physicians, assuming that physician visits took place twice a year preoperatively, 6 times a year the first postoperative year, twice a year the second postoperative year, once a year during the third to fifth year, and once every 2 years thereafter. The proportions of follow-ups performed in public clinics, private offices, and hospitals were extracted from a previous analysis based on RAMQ data.27 Physician visits were distinguished by type: regular or control. Regular visits were spaced at least 4 months apart, whereas control visits took place between 2 regular visits. Patient follow-up associated with PK was identical to patient follow-up associated with PLK. Costs related to postoperative complications included those associated with supplemental physician visits, medications, and medical interventions. Management of postsurgical complications except for glaucoma was determined by expert opinion and total health expenditures were estimated using

Table 1—Base-case transition probabilities relative to postoperative time (PK vs PLK) Postoperative time, y PK

PLK

1

5

10

15

20

25

1

5

10

15

20

25

Waiting for corneal transplant RWaiting for corneal transplant

0.532

0.574









0.487

0.561









Waiting for corneal transplantRSurviving graft with or without complication

0.410

0.115









0.482

0.134









Waiting for corneal transplantRIrreversible graft failure

0.035

0.010









0.007

0.003









Waiting for corneal transplant R Non-eligibility

0.014

0.288









0.014

0.288









Waiting for corneal transplant R Death

0.009

0.014









0.009

0.014









Surviving graft with or without complication R Surviving graft with or without complication



0.964

0.967

0.957

0.935

0.893



0.964

0.967

0.957

0.935

0.893

Surviving graft with or without complication R Irreversible graft failure



0.022

0.011

0.007

0.005

0.004



0.023

0.011

0.007

0.005

0.004

Surviving graft with or without complication R Death



0.014

0.022

0.036

0.060

0.103



0.014

0.022

0.036

0.060

0.103

Irreversible graft failure R Irreversible graft failure



0.986

0.978

0.964

0.940

0.897



0.986

0.978

0.964

0.940

0.897

Irreversible graft failure R Death



0.014

0.022

0.036

0.060

0.103



0.014

0.022

0.036

0.060

0.103

Noneligibility R Noneligibility



0.986

0.978

0.964

0.940

0.897



0.986

0.978

0.964

0.940

0.897

Noneligibility R Death



0.014

0.022

0.036

0.060

0.103



0.014

0.022

0.036

0.060

0.103

Death R Death



1.000

1.000

1.000

1.000

1.000



1.000

1.000

1.000

1.000

1.000

Note: This table only shows the transition probabilities for every 5 years even if specific transition probabilities for each year were used in the model. PK, penetrating keratoplasty; PLK, posterior lamellar keratoplasty.

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

245

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. RAMQ reimbursement data. Annual costs related to glaucoma medications were estimated using data on medication use from the RAMQ drug program database for the period from July 2006 to June 2007. According to expert opinion, approximately 25% of patients experiencing immunological graft rejection will receive cyclosporine for the rest of their lives. For these patients, the lifelong costs associated with Table 2—Base-case probabilities associated with postoperative complications PK

PLK

Graft rejection

0.157

0.072

Glaucoma

0.147

0.019

Bacterial keratitis

0.067

0.000

Endophthalmitis

0.006

0.000

Disc dislocation

0.000

0.079

Note: PK, penetrating keratoplasty; PLK, posterior lamellar keratoplasty.

Table 3—Base-case costs for Markov model (costs/patient) Parameter

Base-case costs, 2007 $

References

Costs associated with corneal transplant Hospital costs

1942



Ophthalmologist fees

550

31

Anaesthesiologist fees

115

31

Ophthalmic antibiotics (pms-erythromycin, oph. ointment, 3.5 g)

24

32

Corticosteroids (ratio-prednisolone 1%, oph. suspension)

248

32

48 750



Maintenance medication

Moria surgical instruments* Costs associated with patient follow-up Preoperative years

68

31

1st postoperative year

161

31

cyclosporine were taken into account. Costs related to postoperative glaucoma were also estimated over time, because this postsurgical complication implies long-term management in terms of pharmacotherapy and patient follow-up. Utility Because most ophthalmic disorders are known to affect the patient’s health-related quality of life and utility, QALYs were used to assess the global impact of both techniques. Utility, varying from 1.0 (perfect health) to 0.0 (death), was estimated using the Brown and Sharma conversion chart. Converting patient visual acuity in the better-seeing eye into utility using the time trade-off method has been validated previously.33,34 In the present analysis, preoperative and postoperative visual acuities associated with PK and PLK were obtained from the literature and were used to estimate patient utilities. More specifically, the average PK-related visual acuities in the better-seeing eye used in the present analysis ranged from 20/200 before surgery to 20/60 12 months after surgery, corresponding to utility values of 0.66 and 0.75, respectively.35 According to a comparative study on DLEK, DSEK, and DSAEK, the mean preoperative and 12-month postoperative visual acuities in the better-seeing eye for these 3 techniques were also estimated at approximately 20/200 and 20/60, corresponding as well to a utility of 0.66 and 0.75, respectively.23 PK- and PLK-related preoperative utilities were used as the utilities associated with the waiting-for-transplant health state, whereas postoperative utilities associated with PK and PLK were used as utilities related to the surviving-graft state. Utilities associated with irreversible graft failure and noneligibility were assumed to be similar to that associated with the waiting-for-transplant state. In parallel, the impact of immunological graft rejection on patient utility was also considered, because this complication is associated with long-term medication use and anxiety related to the high risk of developing irreversible graft failure. Utilities for each Markov state are presented in Table 4.

2nd postoperative year

68

31

3rd–5th postoperative years

34

31

.5th

17

31

Corticosteroids (ratio-prednisolone 1%, oph. suspension)

125

32

Corticosteroids (dexamethasone, oph. ointment, 3.5 g)

33

32

Cyclosporine, magistral, drops

420



Additional physician visits

78

31

Annual medication costs

281

RAMQ database

Additional physician visits (3rd–5th postoperative years)

103

31

Additional physician visits (.5th postoperative year)

51

31

Ophthalmic antibiotics (ofloxacin 0.3%, oph. solution)

19

32

Additional physician visits

59

31

280

31

Table 4—Base-case utility according to health state of the Markov model

postoperative year

Costs associated with postoperative complications Immunological graft rejection

Glaucoma

Microbial keratitis

Endophthalmitis Vitrectomy

Cost-utility Incremental cost-utility ratio (ICUR) was calculated using the following equation: . ICUR~DC=DE~ðCðLK Þ{CðPKÞÞ (EðPLKÞ{EðPK ÞÞ

Tobramicin + Cefazolin

8

32

Markov state

Ophthalmic antibiotics (Ciprofloxacin chlorhydrate, co, 500 mg)

25

32

Waiting for transplant

Additional physician visits

118

31

260

31

Corneal disc dislocation Secondary closing of the operative wound

*The costs of the Moria surgical instrument were amortized according to the number of corneal cuts performed by one Moria surgical instrument, corresponding to a total cost of $97.50 per patient. Note: oph, ophthalmic; co, tablets.

246

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

Utility 0.66

Surviving graft with or without complications

0.75

Immunological graft rejection

0.66

Irreversible graft failure

0.66

Noneligibility

0.66

Death

0.00

Note: Data from the studies by Bahar et al.23 and Beckingsale et al.35 were used to calculate utility values.

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. where DC is the difference between costs associated with PLK (CPLK) and costs associated with PK only (CPK), and DE is the difference between the number of QALYs associated with PLK (EPLK) and the number of QALYs associated with PK (EPK). Costs and QALYs estimated beyond the first year were actualized using a 5% discount rate, as recommended by the Canadian Agency for Drugs and Technologies in Health.36

Table 5—Parameters submitted to deterministic and probabilistic sensitivity analyses Base-case scenario

Min

Max

Hospital costs

1942

1456

2428

Ophthalmologist fees

550

412

688

Anaesthesiologist fees

115

86

143

Maintenance medication (total costs)

272

204

340

48 750

36 562

60 938

Parameter Costs associated with corneal transplant*{, $

Moria surgical instruments

Sensitivity analyses Sensitivity analyses were performed to confirm the robustness of the base-case results. Variations in key parameters and assumptions were carried out in deterministic and probabilistic sensitivity analyses (Table 5). In the deterministic analysis, the impact of input uncertainty was assessed by varying each key parameter individually between upper and lower bounds. Parameters submitted to deterministic analyses were corneal transplant– related costs, utility, discount rates, and incidence of postoperative complications. Assumptions concerning the impact of the advent of PLK on graft accessibility, the proportion of patients with KC in a cohort of patients waiting for transplant, and the number of corneal cuts performed by one Moria surgical instrument were also assessed in a deterministic analysis. Parameter and assumption uncertainties were also evaluated using a probabilistic sensitivity analysis. Simultaneous variations in model inputs were performed using Monte Carlo simulations. Triangular distributions were used for all model inputs submitted to probabilistic sensitivity analysis, except for the impact of PLK on graft accessibility and the number of corneal cuts performed by one Moria surgical instrument, for which a uniform distribution was applied. Results of Monte Carlo simulations were presented as an acceptability curve. The Markov model was constructed using Office Excel 2007 software (Microsoft Corp, Redmond, Wash.) and sensitivity analyses were performed with Crystal Ball software, v. 11.1 (Oracle Corp, Redwood Shores, Calif.). RESULTS

Costs associated with patient follow-up*{, $ Preoperative years

68

51

85

1st postoperative year

161

121

201

2nd postoperative year

68

51

85

3rd–5th postoperative year

34

26

43

.5th

17

13

21

Corticosteroids (ratio-prednisolone 1%, oph. suspension)

125

94

156

Corticosteroids (Dexamethasone, oph. ointment, 3.5g)

33

24

41

Cyclosporine, marginal, drops

420

315

525

Additional physician visits

78

59

98

Annual medication costs

281

211

352

Additional physician visits (3rd–5th postoperative year)

103

77

128

Additional physician visits (.5th postoperative year)

51

38

64

Ophthalmic antibiotics (ofloxacin 0.3%, oph. solution)

19

14

23

Additional physician visits

59

44

73

350

postoperative year

Costs associated with postoperative complications: immunological graft rejection*{, $

Costs associated with postoperative complications: glaucoma*{, $

Costs associated with postoperative complications: microbial keratitis*{, $

Costs associated with postoperative complications: endophthalmitis*{, $ Vitrectomy

280

210

Tobramicin + Cefazolin

8

6

10

Ophthalmic antibiotics (Ciprofloxacin chlorohydrate, co, 500 mg )

25

19

31

Additional physician visits

118

88

147

260

195

325

Waiting for transplant*{

0.66

0.50

0.82

Surviving graft with or without complications*{

0.75

0.56

0.94

Immunological graft rejection*{

0.66

0.50

0.82

Irreversible graft failure*{

0.66

0.50

0.82

Noneligibility*{

0.66

0.50

0.82

Costs associated with postoperative complications: corneal disc dislocation*{, $ Secondary closing of the operative wound Utility values

Postoperative complications (%)*

Base-case results Over a lifetime horizon, PLK was shown to be less costly ($375 915/100 patients) compared with only PK ($444 707/100 patients). The difference in total costs resulted mainly from the better clinical outcomes associated with PLK, because irreversible graft failures and postoperative complications, both less frequent with this surgical technique, are related to higher health-care expenditures. Furthermore, PLK had a positive impact on patient utility, providing more QALYs (965 QALYs/100 patients) than PK (952 QALYs/100 patients). The gain in QALYs associated with PLK resulted from reduced wait time for transplant and a lower incidence of irreversible graft failure and postoperative complications, all related to

Post-PK graft rejection

0.157

0.148

0.166

Post-PK glaucoma

0.147

0.136

0.158

Post-PK bacterial keratitis

0.067

0.061

0.074

Post-PK endophthalmitis

0.006

0.0055

0.007

Post-PLK graft rejection

0.072

0.058

0.085

Post-PLK glaucoma

0.019

0.002

0.035

Post-PLK disc dislocation

0.079

0.063

0.094

Other study parameters Discount rate{

0.05

0

0.03

0.0934

0

0.32

Impact of PLK on graft accessibility*

1.1

1

1.5

Number of corneal cuts performed by one Moria surgical instrument*

500

250

750

% of patients with keratoconus{

*Parameters submitted to both deterministic and probabilistic sensitivity analyses; the interval tested was the same for both analyses. { The interval tested for these parameters was ¡25%. { Parameters submitted to deterministic analyses only. Note: Min, minimum; Max, maximum; oph, ophthalmic; co, tablets; PLK, posterior lamellar keratoplasty; PK, penetrating keratoplasty.

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

247

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. lower utility. Consequently, as presented in Table 6, PLK proved to be less costly and more effective compared with PK only, making it a dominant strategy for the management of patients waiting for a corneal graft. Sensitivity analyses Deterministic and probabilistic sensitivity analyses confirmed the robustness of the base-case results. Compared with PK only, PLK in the management of patients waiting for a corneal graft remained a dominant alternative when most key parameters and assumptions were submitted to deterministic analyses. However, when the utility values associated with surviving graft with or without complications as well as those associated with immunological graft failure/irreversible graft rejection were submitted to deterministic analyses, the dominant relationship of PLK over PK resulted in making PK a more favorable alternative in terms of QALY gain. However, because costs related to the advent of PLK remained lower than those related to PK in the deterministic analysis, PLK never became a dominated strategy. As demonstrated by the acceptability curve (Fig. 2), the probability of PLK being cost effective was 93.3% when a willingness-to-pay threshold of $10 000/QALY was adopted. According to the acceptability curve, PLK represents a dominant alternative in approximately 85% of cases. CONCLUSIONS

According to the results of this economic analysis, the addition of PLK to the management of patients waiting for corneal graft is beneficial. Compared with PK only, PLK proved to be a dominant alternative, because it provided more QALYs and entailed fewer costs. In addition to its Table 6—Incremental cost-utility ratio of PLK compared with PK (per 100 patients) Surgical corneal graft technique

Costs, $

D Costs

QALYs

D QALYs

PK

444 707



952



PLK

375 915

268 792

965

13

ICUR — 25352 Dominant

Note: PLK, posterior lamellar keratoplasty; PK, penetrating keratoplasty; QALY, qualityadjusted life-year; ICUR, cost-utility ratio.

Fig. 2—Results of the probabilistic sensitivity analysis. (PLK, posterior lamellar keratoplasty; PK, penetrating keratoplasty.)

248

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

beneficial impact on costs and patient utility, PLK could improve accessibility to corneal graft, because it is associated with fewer exclusion criteria for eye donation. However, PLK is limited to endothelial corneal conditions and is associated with additional postoperative complications such as corneal disc dislocation, a complication related to significant costs. The present study was the first economic analysis to assess the cost-utility of PLK in the treatment of corneal endothelial diseases. To date, no economic evaluation of PLK has been reported in the literature. However, 2 economic analyses of PK have been conducted. In 2006, a German study by Hirneiss et al.37 reported the costs and utility associated with PK in patients with poor binocular visual acuity. More recently, Roe et al.38 assessed, over a lifetime period, the cost-utility of PK for patients with severe KC. These 2 studies differ in certain respects from the present cost-utility analysis of PLK. First, the present analysis did not specifically consider postoperative astigmatism, contrary to the study by Roe et al.38 Many authors have suggested that PLK is associated with lower rates of postoperative astigmatism compared with PK.23,24 Astigmatism is known to affect visual acuity and consequently, patient utility. Because the outcome measure of the present analysis was based on visual acuities and utility, the clinical impact of astigmatism was indirectly taken into account. However, the costs associated with the treatment of astigmatism (eyeglasses, contact lenses, or keratotomy) were not included in this analysis. Considering that astigmatism is less frequent with PLK, the inclusion of these costs would have accentuated the established dominant relationship. Second, the costs and utilities found in the present study diverge slightly from those found in previous studies. Indeed, Hirneiss et al.37 showed that compared with no treatment, the incremental costs and QALYs associated with PK, estimated for a 10-year timeframe, were J7210 ($11 385) and 0.755 QALYs, respectively. The economic analysis by Roe et al.36 indicated that the incremental costs and QALYs associated with PK, when compared with no treatment, were estimated at US$5934 ($7337) and 5.36 QALYs over a lifetime period. In the present analysis, PLK was associated with fewer costs (2$688) and greater QALYs (+0.13 QALYs) compared with PK. The differences between the incremental costs and utilities found in these economic analyses may first be explained by differences in patient populations and in health-care settings between countries, which greatly affect total health-care expenses. Moreover, in the present study, PLK was compared with another corneal transplant technique, whereas in the 2 above-mentioned studies, PK was compared with no surgery. In this analysis, Brown and Sharma’s conversion chart was used to estimate patient utility associated with the different health states of the model.33,34 This conversion chart, commonly used to evaluate patient utilities associated with different eye disorders or procedures, has been

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. validated previously. Visual acuities found in the studies by Beckingsale et al.35 and Bahar et al.23 were based on large and representative patient samples, allowing a realistic estimation of patient utilities. However, there may be some uncertainty surrounding utility values because the conversion chart represents an indirect measure of patient utility. Deterministic sensitivity analyses revealed that PLK remained a dominant alternative, even when a variation of ¡25% was applied to most utility values. The interval tested in the sensitivity analyses (¡25%) was assumed to be large enough to capture the impact of uncertainty on utility values. Moreover, the postoperative visual acuities used to estimate patient utility in this analysis were those related to the 3 PLK techniques (DLEK, DSEK, and DSAEK). Because newer techniques such as DSAEK may be associated with better postoperative visual acuities, these techniques in the management of patients waiting for corneal graft could be even more advantageous in terms of patient visual acuity and utility. To validate the utility values used in the present analysis, a complementary analysis was performed in which utility was estimated using the EuroQol (EQ-5D) questionnaire. This validated questionnaire was administered to a sample of 15 patients who received PK or PLK at Maisonneuve-Rosemont Hospital. The utilities estimated from this patient sample are presented in Table 7. When these utility values were used in the model, PLK remained a dominant alternative. Because the EQ-5D questionnaire represents an instrument of choice for measuring patient utility according to Canadian Agency for Drugs and Technologies in Health guidelines, the use of this standardized questionnaire for estimating patient utility in the base-case analysis would have been preferable. However, considering that the EQ-5D was administered to a small sample of patients, the estimation of patient utility using Brown and Sharma’s conversion chart was judged more appropriate. In future studies, it would be of interest to estimate patient utility using the EQ-5D questionnaire on a larger sample of patients. The present study has several strengths. First, the extensive sensitivity analyses confirmed the robustness of the base-case results. Moreover, accessibility to corneal graft was considered, which allowed a better understanding of the real impact of PLK in Canadian medical practice. Furthermore, the effectiveness assessment of surgical techniques using patient utility instead of a clinical parameter Table 7—PK-related and PLK-related utilities according to health state of the Markov model, estimation from the EQ-5D questionnaire Markov state

Utility

Waiting for transplant

0.720

Surviving graft with or without complications

0.907

Immunological graft rejection

0.720

Irreversible graft failure

0.720

Noneligibility

0.720

Death

0.000

Note: PK, penetrating keratoplasty; PLK, posterior lamellar keratoplasty; EQ-5D, EuroQol.

is relevant, given the substantial impact of poor visual acuity on patient utility. In addition, the noninclusion of KC patients in the PLK group, considering that they would receive PK even though PLK was available, better reflected clinical practices because they are ineligible for PLK. Nevertheless, to better estimate the real economic impact of the addition of PLK to the management of patients waiting for corneal graft, patients with other nonendothelial diagnoses such as corneal scars, infective keratitis, or chemical burns should have been analyzed similarly to KC patients, because they are also not candidates for PLK. However, because these preoperative diagnoses are seldom observed in overall published cohorts, their clinical and economic impact was considered minimal. Nevertheless, this study has some limitations. First, except for the Moria surgical instrument, the costs associated with PLK equipment (diamond blades, disposal cutters, etc.) were not specifically captured because, on the one hand, the PK-related costs of blades and other equipment were already included in the total hospital costs, and on the other hand, the hospital costs associated with both surgical techniques were considered to be the same. The extra costs of PLK equipment compared with PK equipment were therefore not included in this analysis, except for the Moria surgical instrument. Including the costs of PLK equipment in addition to hospital costs would have resulted in double counting. Moreover, the costs of patient follow-up associated with PK were identical to the costs of patient follow-up associated with PLK; thus, the costs related to the extra time needed in follow-up for PK were not included. The costs associated with patient loss of productivity were also not included, and consequently, the impact of PLK from a societal perspective was not assessed. In addition, this analysis did not take into account the learning curve associated with PLK, because the clinical outcomes resulted from published studies conducted mostly by experienced surgeons. Therefore, the costs related to graft failure and dislocation may have been underestimated. Moreover, the costs related to eye donation were not included in the analysis, because they were assumed to be the same for both surgical techniques. If the present analysis had considered that PLK was performed using precut corneas, the PLK costs related to eye donation would have increased significantly. However, the impact of such a consideration would be minimized from the health-care system perspective adopted in this study, because the hospital costs related to Moria surgical instruments would have been transferred into eye bank costs. Finally, long-term data on PLK were not available at the time of the analysis, which limited the estimation over time of postoperative-related outcomes. In conclusion, although PLK can only be used for the treatment of corneal endothelial diseases, the results of this study indicate that it would be beneficial from a costeffectiveness standpoint in the management of patients waiting for corneal graft. CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

249

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. This study was supported by the Canadian Institutes of Health Research (CIHR-RFM-7256) (I. Brunette and J. Lachaine). Isabelle Brunette was the recipient of the Charles-Albert Poissant Research Chair in Corneal Transplantation, University of Montreal, Montreal, Que. The authors would like to acknowledge the significant contributions of Dr. Michelle Mabon, Dr. Michel Le Franc¸ois, Solmaz Moghadaszadeh, and MarieEve Choronzey. The authors have no proprietary or commercial interest in any materials discussed in this article.

18.

19.

20.

REFERENCES 1. World Health Organization. Human organ and tissue transplantation. Available at: http://www.who.int/ethics/topics/ human_transplant/en/. Accessed November 8, 2007. 2. Fishman JA, Greenwald MA, Kuehnert MJ. Enhancing transplant safety: a new era in the microbiologic evaluation of organ donors? Am J Transplant 2007;7:2652–4. 3. Eye Bank Association of America. About us: frequently asked questions. Available at: http://www.restoresight.org/aboutus/ faqs.htm#3. Accessed November 8, 2007. 4. Eye Bank Association of America. Press release; Eye bank association of america announces eye banking statistics for 2008. Available at: https://www.restoresight.org/files/ 2008pressrelease_statreport.pdf. Accessed Apr. 28, 2009. 5. McCready W. The occasional eye removal for corneal transplantation. Can J Rural Med 2008;13:189–93. 6. Thompson RW Jr, Price MO, Bowers PJ, Price FW Jr. Longterm graft survival after penetrating keratoplasty. Ophthalmology 2003;110:1396–402. 7. Cao KY, Dorrepaal SJ, Seamone C, Slomovic AR. Demographics of corneal transplantation in Canada in 2004. Can J Ophthalmol 2006;41:688–92. 8. Rasouli M, Caraiscos VB, Slomovic AR. Efficacy of Routine Notification and Request on reducing corneal transplantation wait times in Canada. Can J Ophthalmol 2009;44:31–5. 9. Boisjoly H, Gresset J, Charest M, et al. The VF-14 index of visual function in recipients of a corneal graft: a 2-year followup study. Am J Ophthalmol 2002;134:166–71. 10. Goins KM. Surgical alternatives to penetrating keratoplasty II: endothelial keratoplasty. Int Ophthalmol 2008;28:233–46. 11. Melles GR, Lander F, Beekhuis WH, Remeijer L, Binder PS. Posterior lamellar keratoplasty for a case of pseudophakic bullous keratopathy. Am J Ophthalmol 1999;127:340–1. 12. Melles GR, Lander F, van Dooren BT, Pels E, Beekhuis WH. Preliminary clinical results of posterior lamellar keratoplasty through a sclerocorneal pocket incision. Ophthalmology 2000; 107:1850–6. 13. Tan DT, Mehta JS. Future directions in lamellar corneal transplantation. Cornea 2007;26:S21–S28. 14. Melles GR, Wijdh RH, Nieuwendaal CP. A technique to excise the descemet membrane from a recipient cornea (descemetorhexis). Cornea 2004;23:286–8. 15. Ousley PJ, Terry MA. Stability of vision, topography, and endothelial cell density from 1 year to 2 years after deep lamellar endothelial keratoplasty surgery. Ophthalmology 2005;112:50–7. 16. Terry MA, Ousley PJ. Deep lamellar endothelial keratoplasty in the first United States patients: early clinical results. Cornea 2001;20:239–43. 17. Terry MA, Ousley PJ. Replacing the endothelium without corneal surface incisions or sutures: the first United States

250

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31. 32. 33. 34.

35.

clinical series using the deep lamellar endothelial keratoplasty procedure. Ophthalmology 2003;110:755–64. Terry MA, Ousley PJ. Rapid visual rehabilitation after endothelial transplants with deep lamellar endothelial keratoplasty (DLEK). Cornea 2004;23:143–53. Terry MA, Ousley PJ. Deep lamellar endothelial keratoplasty visual acuity, astigmatism, and endothelial survival in a large prospective series. Ophthalmology 2005;112:1541–8. Terry MA, Ousley PJ. Deep lamellar endothelial keratoplasty: early complications and their management. Cornea 2006;25:37–43. Terry MA. Endothelial keratoplasty: clinical outcomes in the two years following deep lamellar endothelial keratoplasty (an American Ophthalmological Society thesis). Trans Am Ophthalmol Soc 2007;105:530–63. Terry MA, Wall JM, Hoar KL, Ousley PJ. A prospective study of endothelial cell loss during the 2 years after deep lamellar endothelial keratoplasty. Ophthalmology 2007;114:631–9. Bahar I, Kaiserman I, McAllum P, Slomovic A, Rootman D. Comparison of posterior lamellar keratoplasty techniques to penetrating keratoplasty. Ophthalmology 2008;115:1525–33. Heidemann DG, Dunn SP, Chow CY. Comparison of deep lamellar endothelial keratoplasty and penetrating keratoplasty in patients with Fuchs endothelial dystrophy. Cornea 2008;27:161–7. Armour RL, Ousley PJ, Wall J, Hoar K, Stoeger C, Terry MA. Endothelial keratoplasty using donor tissue not suitable for fullthickness penetrating keratoplasty. Cornea 2007;26:515–9. Phillips PM, Terry MA, Shamie N, et al. Descemet’s stripping automated endothelial keratoplasty (DSAEK) using corneal donor tissue not acceptable for use in penetrating keratoplasty as a result of anterior stromal scars, pterygia, and previous corneal refractive surgical procedures. Cornea 2009;28:871–6. Roussy JP, Aubin MJ, Brunette I, Lachaine J. Cost of corneal transplantation for the Quebec health care system. Can J Ophthalmol 2009;44:36–41. Ghosheh FR, Cremona F, Ayres BD, et al. Indications for penetrating keratoplasty and associated procedures, 2001– 2005. Eye Contact Lens 2008;34:211–4. Beauchemin C, Brunette I, Lachaine J. Graft survival associated with penetrating keratoplasty and posterior lamellar keratoplasty. Poster presentation; presented at the annual meeting of the Canadian Ophthalmological Society in June 2009, Toronto, Ont. Statistics Canada. Complete life table, Canada, 2000 to 2002. Available at: http://www.statcan.gc.ca/pub/84-537-x/4064441-eng. htm. Accessed Mar. 24, 2009. Re´gie de l’assurance maladie du Que´bec. Manuel des me´decins spe´cialistes. Service des relations avec la cliente`le; Que´bec, 2008. Re´gie de l’assurance maladie du Que´bec. Liste des me´dicaments. Service des relations avec la cliente`le; Que´bec, 2008. Brown GC. Vision and quality-of-life. Trans Am Ophthalmol Soc 1999;97:473–511. Brown MM, Brown GC, Sharma S, Smith AF, Landy J. A utility analysis correlation with visual acuity: methodologies and vision in the better and poorer eyes. Int Ophthalmol 2001;24:123–7. Beckingsale P, Mavrikakis I, Al-Yousuf N, Mavrikakis E, Daya SM. Penetrating keratoplasty: outcomes from a corneal unit compared to national data. Br J Ophthalmol 2006;90:728–31.

Economic impact of posterior lamellar keratoplasty—Beauchemin et al. 36. Guidelines for the economic evaluation of health technologies: Canada [3rd Edition]. Ottawa: Canadian agency for drugs and technologies in health;2006. 37. Hirneiss C, Neubauer AS, Niedermeier A, Messmer EM, Ulbig M, Kampik A. Cost utility for penetrating keratoplasty in patients with poor binocular vision. Ophthalmology 2006; 113:2176–80.

38. Roe RH, Lass JH, Brown GC, Brown MM. The value-based medicine comparative effectiveness and cost-effectiveness of penetrating keratoplasty for keratoconus. Cornea 2008;27:1001–7. Keywords: corneal transplantation, corneal endothelium, penetreting keratoplasty, posterior lamellar keratoplasty, cost utility analysis

CAN J OPHTHALMOL—VOL. 45, NO. 3, 2010

251