coste eficacia master!!!

JOURNAL CLUB

CLINICAL SCIENCES

Cost-effectiveness of 2 Approaches to ManagingNasolacrimal Duct Obstruction in Infants

The Importance of the Spontaneous Resolution Rate

Kevin D. Frick, PhD; Luxme Hariharan, MD, MPH; Michael X. Repka, MD, MBA;Danielle Chandler, MSPH; B. Michele Melia, ScM; Roy W. Beck, MD, PhD;for the Pediatric Eye Disease Investigator Group (PEDIG)

Objective: To assess the impact of the rate of sponta-neous resolution of congenital nasolacrimal duct ob-struction on the relative cost-effectiveness of deferred na-solacrimal duct probing in a surgical facility (DFPS)compared with an immediate office-based probing sur-gery (IOPS).

Methods: Data from the literature, Medicare 2009 feeschedule, and consensus assumptions were combined topopulate a model of outcomes of 2 treatment strategies:immediate office-based probing (IOPS) and deferredfacility-based probing (DFPS) (deferred for 6 months).Sensitivity analyses were conducted, varying the 6-monthspontaneous resolution rate from 50% to 90%. Addi-tional factors varied during analyses included surgical costand each procedures probability of success. Outcomesmeasured were overall cost of treatment, chance of cure,and months of symptoms avoided by 18 months of life.

Results: Under the base case, assuming a 75% spontane-ous resolution rate during 6months prior to deferred prob-ing, IOPS is more expensive ($771 vs $641) and slightlyless effective (93.0% vs 97.5%) thanDFPS, although IOPScosts only $44 per month of symptoms avoided. At spon-taneous resolution rates between 50%and68%, IOPS costsless than DFPS (from $2 to $342 less), although it also isslightly less effective (from 2.0% to 3.8% less). At a 90%spontaneous resolution rate, IOPS costs $169 per monthof symptoms avoided. As the rate of spontaneous resolu-tion falls, the cost per additional success for DFPS in-creases to $16 709 at a 50% spontaneous resolution rate.

Conclusion: The relative cost-effectiveness of these strat-egies for treatment of nasolacrimal duct obstruction de-pends on the spontaneous resolution rate after diagnosis.

Arch Ophthalmol. 2011;129(5):603-609

C ONGENITAL NASOLACRI-mal duct obstruction(NLDO) is a commoncondition in the first yearof life. Many cases re-solve spontaneously or with massage by12months of age.1-4 For children youngerthan 6 months, nonsurgical treatment isusually administered, including antibi-otic eyedrops and massage of the lacri-mal sac. For children older than 6months,some surgeons offer probing of the naso-lacrimal duct to clear the blockage in theoffice settingwith topical anesthesia.Othersurgeons continue medical managementfor varying lengths of time, with subse-quent probing under general anesthesia ina hospital outpatient department (HOPD)or ambulatory surgical center (ASC)if spontaneous resolution has not oc-curred. The advantages of early probingin the office setting are avoidance of gen-eral anesthesia, immediate resolution ofsymptoms, fewer physician visits, fewer an-tibiotic prescriptions, and less costly pro-

cedures. The advantages of deferring theprobing include more comfort with theprocedure for the infant and avoidance ofa surgical procedure and its associatedcosts if spontaneous resolution occurs.

Controversy remains among surgeonsas to the preferred approach. Data that as-sist in the decision-making process in-clude the rate of spontaneous resolutionafter the age at which an office proceduremight be done and the success rates forprobing procedures done early in theoffice and in a surgical facility after a pe-riod of observation. Awide range of spon-taneous resolution rates has been docu-mented2,4,5 (from 32% to 95%), althoughthe spontaneous resolution rate of symp-toms present in infants at 6 months ofage has been addressed in only 1 report,6

a retrospective study that found the rateto be 70% by 12 months of age (26 of 37

Journal Club slides availableat www.archophthalmol.comAuthor Affiliations:

Department of Health Policyand Management, JohnsHopkins Bloomberg School ofPublic Health (Drs Frick andHariharan), and WilmerOphthalmological Institute,The Johns Hopkins University(Dr Repka), Baltimore,Maryland; and Jaeb Center forHealth Research, Tampa,Florida (Mss Chandler andMelia and Dr Beck).Group Information: A list ofthe members of the PEDIG waspublished in Ophthalmology.2008;115(3):584, e2-e3;J AAPOS. 2009;13(3):306-307;and Arch Ophthalmol.2009;127(5):636-637.

ARCH OPHTHALMOL /VOL 129 (NO. 5), MAY 2011 WWW.ARCHOPHTHALMOL.COM603

2011 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ on 05/30/2013

infants). The overall probability of success for nasolac-rimal duct probing was determined in a prospective ob-servational study7 to be 78%; however, there was a re-duced probability of success when the procedure wasperformed in the office (72%) vs in a surgical facility(80%).

The purpose of this analysis was to assess the impactof the rate of spontaneous resolution of congenital NLDOon the relative cost-effectiveness of deferred nasolacri-mal duct probing surgery in a facility (DFPS) comparedwith immediate office-based probing surgery (IOPS).

METHODS

THE MODEL

A decision tree was used to compare the costs and outcomesfor 2 common options for treating NLDO in children evalu-ated between 6 and 9months of age: IOPS and DFPS if the con-dition had not spontaneously resolved within 6months of pre-sentation. These 2 options are shown as choices at the initialnode in the decision tree (Figure 1). Subsequent to the choiceof the initial approach, we determined potential next steps inthe management of NLDO after the initial surgery. For IOPS,the first options are success or failure of the office probing. Forinitial treatment failures, the options include spontaneous reso-lution occurring prior to another surgery, a second surgery per-formed in the office, a second surgery performed in a facility(probing, balloon catheter dilation, or nasolacrimal duct intu-bation), and no further treatment before the child is 18monthsold. The secondary surgerymay also succeed or fail. In ourmodelthere is no provision for a third surgical procedure; therefore,failure of the second procedure is assumed to be a failure ofthe entire treatment regimen.

The DFPS option is more complex. The primary motiva-tion for considering DFPS is the chance for spontaneous reso-

lution, thus avoiding the need for surgery and its associatedhealth care costs. With higher 6-month rates of spontaneousresolution, a smaller number of infants will need to undergosurgical probing. Among those whose NLDO does not resolvewithin 6 months, some parents may decide not to pursue sur-gical intervention. A subset of those cases will resolve with notreatment, and others will be considered treatment failures.Among children who undergo deferred probing in a facility,the procedurewill either succeed or fail, with somewhose treat-ment failed undergoing a second surgery. The second surgeryin a facility (probing, balloon catheter dilation, or nasolacri-mal intubation) also can result in either success or failure. Thereis no provision for a third surgical procedure; therefore, fail-ure of the second procedure is assumed to be a failure of theentire treatment regimen.

FACTORS: COSTS AND PROBABILITIES

The point estimates for costs and probabilities are shown inTable 1. Data on costs were obtained primarily from the 2009Medicare Fee Schedule.8,9 We assumed full phase-in of ambu-latory surgery rates at 59% of the hospital outpatient opera-tions schedule. The cost of a facility surgery as a follow-up toa failed surgery (either an office probing or a facility probing)is based on the 2009 Medicare Fee Schedule costs,8,9 assumingthat the distribution of type of the secondary surgery wouldapproximate that found in a prospective observational study10,11:balloon catheter dilation in 43% of cases, nasolacrimal duct in-tubation in 45%, and a second probing procedure in 12%.

A 2008 study7 of NLDO treatment by the Pediatric Eye Dis-ease Investigator Group (PEDIG) was themain source of prob-ability data on clinical outcomes. The probability of success foroffice-based probing was 75%, based on unpublished data onpatients reported in that study but limited to a subset of 98 pa-tients who received office probing between 6 and 9 months ofage. The probability of success for a facility-based probing was80%, based on unpublished data on patients reported in the

NLDOtreatmentstrategy

IOPS

Success

Spontaneous resolutionbefore age 18 mo

No spontaneous resolutionbefore age 18 mo

Spontaneous resolution before retreatment

OPS retreatment before age 18 mo

FPS retreatment before age 18 mo

Failure and no retreatment before age 18 mo

Spontaneous resolution before planned DFPS

Planned DFPS not performed by approximately age 15 mo

FPS by approximately age 15 mo

DFPS

0

0.5

10.5

Calculated

Success

Failure1

10.5Success

Failure1

10.5

8.25

10.5

7

10.5

6

10.5

Success

Success

Failure

FPS retreatmentbefore age18 mo

Failure andno retreatmentbefore age18 mo

Figure 1. Decision tree used in cost-effectiveness analysis. Outcomes shown at terminal nodes are average months experiencing nasolacrimal duct obstruction(NLDO) after presentation between 6 and 9 months of age. DFPS indicates deferred facility probing surgery; failure, presence of at least 1 symptom of NLDO(epiphora, increased tear lake, and mucous discharge); FPS, facility probing surgery; IOPS, immediate office-based probing surgery; OPS, office probing surgery;and success, resolution of NLDO symptoms (ie, absence of epiphora, increased tear lake, and mucous discharge) by age 18 months.




2008 PEDIG study but limited to a subset of 171 patients whoreceived probing in a facility between 12 and 17months of age.The probability of success for a facility-based surgery that wasa follow-up to failed surgerywas 78%, based on the success ratesreported from a prospective observational study10,11 of second-ary procedures (balloon catheter dilation [77%], nasolacrimalduct intubation [84%], and a second probing procedure [56%]),weighted according to the distribution of procedures in the study(balloon catheter dilation [43%], nasolacrimal duct intuba-tion [45%], and a second probing procedure [12%]). The prob-ability of success for an office-based surgery that was a fol-low-up to failed surgery was assumed to be 56%, based on datafrom a prospective study10 of 20 patients. Other estimates in-cluded in Table 1 were made by expert consensus.

OUTCOMES

The outcomes in themodel include an indicator of clinical suc-cess and the number of months of NLDO symptoms. Success

was defined as the absence of epiphora, increased tear lake, andmucus discharge. The number of symptomatic months expe-rienced when reaching each terminal node was calculatedthrough 18 months of age. Months are counted from the timeof the initial office consultation at 6 to 9 months of age (aver-age, 7.5 months) because the focus is placed on the number ofmonths of symptoms experienced subsequent to the choice oftreatment option. Combining the outcome at each end nodewith the probability of reaching each end node allowed us todescribe the expected success rate and calculate the expectednumber of months that NLDO would be present.

For the IOPS option, the number of months with symp-toms depends on the outcome of the immediate office prob-ing. Initial success is consistent with no months of symptomsafter diagnosis. When retreatment is needed, it is assumed tooccur after 1 additionalmonth of symptoms. Success ends symp-toms; cases in which surgery fails a second time are assumednot to resolve, and an average of 10.5months of symptomswould

Table 1. Cost and Probability Factors to Populate the Decision Tree

Factor Type, Factor (Applicable CPT Code) SourceaPoint

Estimateb RangeDistribution

Type

CostInitial physicians office consultation (92004) 8 131 . . . . . .Antibiotic for children receiving deferred surgery (tobramycin sulfate, 0.3%) c 16 . . . . . .Office procedure: professional fee and procedure reimbursement (68810) 8 198 . . . . . .Postprocedure corticosteroidal antibiotic (tobramycin dexamethasone, 0.1%) c 74 . . . . . .Established patient office visit (99213) 8 63 . . . . . .Anesthesia for facility surgery 8 250 . . . . . .Facility surgery professional fee for initial surgery (68811) 9 172 . . . . . .Facility surgery professional fee for secondary surgery (68811, 68815, 68816) 9 208 . . . . . .ASC facility fee for initial surgery 8 607 . . . . . .ASC facility fee for secondary surgery d 681 . . . . . .HOPD facility fee for initial surgery 8 1276 . . . . . .HOPD facility fee for secondary surgery e 1408 . . . . . .

ProbabilitySuccess of IOPS f 0.75 0.66-0.82 PERTSpontaneous resolution after immediate office surgery and prior to retreatment Consensus 0.03 0.02-0.04 PERTFailure to undergo repeat surgery after initial office failure Consensus 0.05 0.03-0.07 PERTSuccess of office surgery after initial office failure 10 0.56 0.33-0.76 PERTRepeat office surgery after initial office failure Consensus 0.10 0.05-0.15 PERTSuccess of repeat surgery after initial surgery failure g 0.78 0.71-0.83 PERTSpontaneous resolution prior to deferred facility surgery Consensus 0.75 0.50-0.90 CalculatedChildren with NLDO resolving before antibiotic prescription renewal Consensus 0.50 0.40-0.60 PERTChildren who reach date of scheduled surgery with a return clinic visit Consensus 0.50 . . . . . .Children not undergoing scheduled deferred surgery Consensus 0.05 0.03-0.07 PERTSpontaneous resolution between 15 and 18 mo Consensus 0.02 0.01-0.03 PERTFacility surgeries at ASC 7 0.32 0.30-1.00 PERTFacility surgeries at HOPD 7 0.68 0.00-1.00 PERTSuccess of initial facility surgery 7 0.80 0.74-0.85 PERTFailure to undergo repeat facility surgery after facility failure Consensus 0.05 0.03-0.07 PERT

Abbreviations: ASC, ambulatory surgical center; CPT, Current Procedural Terminology12; DFPS, deferred facility probing surgical procedure; ellipses, range anddistribution type do not apply to the factor (ie, only the point estimate was evaluated); HOPD, hospital outpatient department; IOPS, immediate office-basedprobing surgical procedure; NLDO, nasolacrimal duct obstruction; PERT, program evaluation and review technique.

aNumbers represent reference numbers.bPoint estimates for costs are given in US dollars.cCost of drug at http://www.drugstore.com, accessed August 8, 2009.dBased on Medicare Physician Fee Schedule 20099 ASC costs estimating that the distribution of type of operative procedure would approximate that found in a

prospective observational study10,11 of secondary procedures: balloon catheter dilation ($780) in 43% of cases, nasolacrimal intubation ($607) in 45%, and asecond probing procedure ($607) in 12%.

eBased on Medicare Physician Fee Schedule 20099 HOPD costs estimating that the distribution of type of surgical procedure would approximate that found in aprospective observational study10,11 of secondary procedures: balloon catheter dilation ($1582, including $306 for cost of balloon catheter) in 43% of cases,nasolacrimal intubation ($1276) in 45%, and a second probing procedure ($1276) in 12%.

fData from an unpublished subset of data published by the PEDIG in 2008.7gBased on the success rates reported in an prospective observational study10,11 of secondary procedures (balloon catheter dilation [77%], nasolacrimal

intubation [84%], and a second probing procedure [56%]), weighted according to the distribution of procedures in the study (43% of patients received ballooncatheter dilation, 45% received nasolacrimal intubation, and 12% received a second probing procedure). Range based on 95% confidence interval based onweighted data.




occur in infants through 18 months of age. This number ofmonths is based on the assumed uniform distribution of pre-sentation between 6 and 9 months of age. If the first treatmentfails but the infants symptoms then resolve spontaneously, theresolution is assumed to be distributed uniformly; thus, the av-erage time to resolution is 0.5 months from the initial surgery.

For the DFPS option, cases that spontaneously resolve be-fore the surgery scheduled after 6 months of observation are as-sumed to resolve with a uniform distribution with an average of3 months of symptoms. Infants successfully treated at the firstsurgery will experience 6 months of symptoms. If NLDO doesnot resolve in the first 6 months and the infants do not undergosurgery, their conditionmay still resolve. This average durationof symptoms is derived from first appointment with the physi-cian between 6 and 9months of age, 6months of symptomspriorto the possible deferred surgery at 12 to 15months of age, and auniform distribution of resolution between 12 to 15 monthsthrough 18 months of age. An infant seen initially at 6 monthsof agewouldhave an expecteddurationof symptomsof 9months.One seen initially at 9 months of age would have an expectedduration of 7.5 months. The midpoint between these 2 dura-tions is 8.25 months. Children whose NLDO does not resolvebetween 15 and 18 months of age will experience an average of10.5months of symptoms.Childrenwith retreatment that is suc-cessful will experience 7months of symptoms. If a second treat-ment is not successful, symptoms will continue for an averageof 10.5 months through 18 months of age.

COST-EFFECTIVENESS ANALYSIS

All costs are reported in 2009 US dollars. The costs used forthe analysis are only the direct medical care costs and do notinclude costs of missed wages or travel. Thus, the perspectiveof this analysis is that of the health care insurer. The base caseincremental cost-effectiveness analysis assumed a spontane-ous resolution rate of 75%. The base case results are reportedboth in terms of overall success at 18 months of age and ofmonths of symptoms avoided at 18months of age. Primary andsensitivity analyses were performed using Microsoft Excel(Microsoft Corporation, Redmond, Washington).

SENSITIVITY ANALYSIS

Three sensitivity analyses (ie, analyses changing assumptionsto determine how critical those assumptions are to determin-ing the result) were conducted. First, we systematically variedthe rate of spontaneous resolution from 50% to 90% whileusing all other factors from the base case. For each spontane-ous resolution proportion, the results indicate which treat-

ment alternative costs less, which has the higher probabilityof success, and the difference in cost to achieve each addi-tional success.

Second, in addition to varying the spontaneous resolutionrate, we varied the relative cost of the immediate office sur-gery and deferred facility surgery. Although changing the costsof both surgical procedures at once would change the cost peradditional success, it would not alter the spontaneous resolu-tion rate at which 1 treatment approach dominates (ie, is bothless expensive andmore effective) because the costs in the cur-rent health care system move in the same direction. Thus, achange in cost would simply expand or contract the incremen-tal cost ratio by the percentage by which the costs are changed.Instead, because the facility fee for performing the surgery inan HOPD is more than double the facility fee for performingthe surgery in an ASC, we varied the relative cost by varyingthe probability of a deferred facility surgery case being per-formed at an ASC from 0% to 100%. Examining the impact ofsite of service for deferred probing results in wide variation inthe relative cost between DFPS and IOPS and allows us to de-termine how the results might change in markets with differ-ent availability of free-standing ASCs.

Third, we performed a probabilistic sensitivity analysis, vary-ing all assumed probabilities (see the ranges defined in Table 1)to explore the effect on the qualitative conclusions. Directmedi-cal costs were held constant in the probabilistic sensitivity analy-sis. For all probabilities except the spontaneous resolution rate,the PERT (program evaluation and review technique, named be-cause of the focus on expert opinion) distribution was used toallow for a minimum (lower bound), a most likely (point esti-mate), and a maximum (upper bound) while having a distribu-tion shape that is not triangular. The distributions of the prob-abilitieswere based on confidence intervals in cases inwhichdatawere available and otherwise on expert consensus. When datafor estimates were derived from multiple sources, we used aweighted average and a confidence interval based on similarweights. A probabilistic sensitivity analysis was performed at thebase spontaneous resolution rate (75%) to explore the impactof changes in these variables. The analysis used 1000 repeatedsimulations to examine the effect of possible random variationin the probabilities of success and appropriate follow-up.

RESULTS

The results of the base case analysis using a spontane-ous rate of resolution of 75% are shown in Table 2. Theaverage cost for the DFPS option is $641, and the aver-age cost of IOPS is $771. Using resolution ofNLDO symp-

Table 2. Base Case Analysisa

Initial Option

Deferred Facility-Based Probing Surgery Immediate Office-Based Probing Surgery

Cost, $b 641 771Incremental cost, $b . . . 130Probability of success, % 97.5 93ICER, $ per successb Dominatesc . .Months of symptoms 3.9 0.9Months of symptoms averted . . . 3ICER, $ per month of symptoms avertedb . . . 44

Abbreviations: ellipses, not applicable; ICER, incremental cost-effectiveness ratio (additional dollars spent per additional unit of outcome).aSeventy-five percent spontaneous resolution rate from enrollment of infants aged 6 to 9 months prior to deferred probing after 6 months of observation.bValues are rounded to the nearest US dollar.cDominates implies that the option is less expensive and more effective.




toms at 18 months of age as the outcome, DFPS wouldhave a probability of success of 97.5%, compared with93.0% for IOPS. The average number ofmonths to symp-tom resolution with IOPS is 0.9, and the average num-ber ofmonths of symptomswithDFPS is 3.9, using a spon-taneous rate of resolution of 75%.Thus, the choice of IOPSavoids an average 3.0 months of symptoms. The addi-tional health care expenditures for IOPS per month ofsymptoms avoided is $44.

Suspecting that the spontaneous resolution rate wouldbe of paramount importance in this analysis, we per-formed sensitivity analyses inwhich the spontaneous reso-lution rate was systematically varied from 50% to 90%(eTable [http://www.archophthalmol.com]). The incre-mental expense with DFPS ranges from a cost of $342per case at 50% spontaneous resolution to savings of $413per case at a 90% spontaneous resolution.

Using themodel, we evaluated the impact of spontane-ous resolution rate on overall proportions of success at 18months of age (eTable). The probability of success ofDFPSincreases from 95% to 99% as the spontaneous resolutionrate increases from50% to 90%.Comparedwith IOPS, theprobability of success ofDFPS is 2%higherwhen the spon-taneous resolution rate is 50% and 6% higher when thespontaneous resolution rate is 90%.Thus,DFPS always hasa higher probability of eventually alleviating the symp-toms before the child is 18 months of age.

The incremental cost-effectiveness ratioof extradollarsspentonDFPSforeachadditionalsuccessfullyresolvedcaseis sensitive to the spontaneous resolution rate. Figure 2shows that at spontaneous resolutionratesof68%ormore,DFPS is both less expensive and more effective (ie, DFPSdominates IOPS). For spontaneous resolution rates of lessthan68%,DFPS ismoreexpensive, althoughslightlymoreeffective.The incremental cost-effectiveness ratio increasesto $16 709 dollars per additional DFPS success when thespontaneous resolution rate is only 50%.

When focusing on months of symptoms averted by18 months of age, IOPS always results in fewer months

of symptoms at spontaneous resolution rates between 50%and 90%. At a spontaneous resolution rate of 50%, IOPSis associated with 3.9 months of symptoms averted by18 months of life; at 90% spontaneous resolution, IOPSis associated with 2.4 months of symptoms averted bythat age. Figure 3 shows the cost per additional monthof symptoms averted by using IOPS rather than DFPS.At the highest spontaneous resolution rate we evaluated(90%, the least favorable for IOPS), the extra cost permonth of symptoms averted by IOPS is $169. At spon-taneous resolution rates between 69% and 90%, IOPS ismore expensive but results in fewer months of symp-toms. At rates of spontaneous resolution less than 69%,IOPS costs less and has fewer months of symptoms, thusdominating DFPS.

In the second sensitivity analysis, changing the rela-tive cost of the 2 surgeries did not result in a substantialchange in the rate of spontaneous resolution at whichDFPS becomes less expensive. In themodel with all basevalues, the threshold is 68%. If all deferred facility prob-ings are assumed to be performed in an ASC, the thresh-old is 64%, whereas if all deferred facility probings areassumed to be performed in an HOPD, the threshold is70%. Thus, changes in the relative cost lead to changesin the threshold that span, at most, a 6-percentage-point range. When all deferred surgeries are performedat ASCs and the spontaneous resolution rate is only 50%,an extra $9145 is spent per extra DFPS success; whenall deferred surgeries are performed in an HOPD, the in-cremental cost-effectiveness ratio is $20 269 per extraDFPS success. In contrast, when the spontaneous reso-lution rate is 90% and all deferred surgeries are per-formed in ASCs, the incremental cost-effectiveness ra-tio is $147 permonth of symptoms averted by IOPS;whenall deferred surgeries are performed in HOPDs, the in-cremental cost-effectiveness ratio is $179 per month ofsymptoms averted by IOPS.

In the probabilistic sensitivity analysis using re-peated sampling from the random distributions of the

18 000

14 000

12 000

16 000

10 000

8000

6000

4000

2000

0

0.50 0.55 0.60 0.65 0.70 0.75

DFPS dominates IOPS

0.80 0.85 0.90Spontaneous Resolution Rate

Incr

emen

tal C

ost-e

ffect

iven

ess

Ratio

, $ p

er E

xtra

Suc

cess

at A

llevi

atin

g Sy

mpt

oms

by 1

8 m

o of

Age

Usi

ng D

FPS

Figure 2. Dollars per additional success by 18 months of age using deferredfacility probing surgery (DFPS) at probabilities of spontaneous resolution inthe 50% to 90% range. IOPS indicates immediate office-based probingsurgery.

180

140

120

160

100

80

60

40

20

0

0.50 0.55 0.60 0.65 0.70 0.75

IOPS dominates DFPS

0.80 0.85 0.90Spontaneous Resolution Rate

Incr

emen

tal C

ost-e

ffect

iven

ess

Ratio

, $ S

pent

per

M

onth

of S

ympt

oms

Aver

ted

by 1

8 m

o of

Age

Usi

ng IO

PS

Figure 3. Dollars per month of symptoms averted by 18 months of age usingimmediate office-based probing surgery (IOPS) at probabilities of spontaneousresolution in the 50% to 90% range. DFPS indicates deferred facility probingsurgery.




probabilities defined by range, type, and data source inTable 1, holding costs constant, assuming a 75% spon-taneous resolution rate, and using success as the out-come of interest, the DFPS is always less expensive andmore likely to succeed. When focusing on months ofsymptoms as the outcome, IOPS is always more expen-sive but results in fewer months of symptoms. The rangeof extra health care spending on IOPS at a 75% sponta-neous resolution rate varies from less than $10 to $102per month of symptoms averted. The values for percen-tiles of 2.5 and 97.5 are $19 and $81, respectively.

COMMENT

There is uncertainty aboutwhether to treat infantileNLDOimmediately in the office setting or to wait for severalmonths for spontaneous resolution and operate if nec-essary in a surgical facility. In this report we have mod-eled the costs and probabilities of success for these 2 treat-ment regimens. The rate of spontaneous resolution is akey factor in predicting the relative cost-effectiveness andclinical outcomes to be expected from these ap-proaches. Most estimates of the rate of spontaneous reso-lution include infants younger than 6 months who areexpected to have a high rate of spontaneous resolutionand in whom no probing would be planned immedi-ately.2,4,5 Few data address the clinically pertinent ques-tion of what to do for the symptomatic infant aged 6 to9 months for whom a surgeon might perform an officeprobing immediately or decide to wait for up to 6monthsand perform a facility probing. Paul6 found that among37 eyes with NLDO symptoms at 6 months of age, 26resolved (70%) without surgical intervention by 12months of age and that among 23 eyes withNLDO symp-toms at 9months, 12 cleared (52%) by 12months of age.

In this cost-effectiveness model we used a 6-monthspontaneous resolution rate of 75% for infants between6 and 9 months of age with symptoms of NLDO whoeither underwent IOPS or waited 6 months for facilitysurgery. Children whose procedures failed could be re-treated once and the outcomes assessed at 18 months ofage. The IOPS andDFPS approaches have nearly the samecost and probability of success; DFPS is $130 less ex-pensive and slightly (4.5%) more successful. However,the DFPS approach is associated with an average of 3.0extra months of symptoms. The average extra cost permonth of symptoms avoidedwhenusing IOPS rather thanDFPS is $44.

Our sensitivity analyses varied the rate of spontane-ous resolution and showedhow critical spontaneous reso-lution is in our model. A difference of as little as 8% inthe spontaneous resolution rate (68% vs 60%)means thedifference between the DFPS approach being more ef-fective and less costly andDFPS costingmore than $5000per expected extra successful treatment. Clinically rel-evant changes in relative costs change the threshold rateof spontaneous resolution above which DFPS becomesless costly andmore effective, although all thresholds arewithin 6 percentage points and are below the base-casespontaneous resolution rate of 75%.More certainty aboutthe costs (as related to the localmarket structure inwhich

clinical decisions are beingmade) would become impor-tant only if future definitive studies reveal a spontane-ous resolution rate lower than 75%. The sensitivity analy-ses on costs also reveal how important it is for stakeholdersto understand the impact of site of service, which sub-stantially changes the cost profile for DFPS bymore thandoubling the cost of the initial surgery when all de-ferred probings are performed in the ASC comparedwiththeHOPD. The changes in relative cost had amuch largereffect on the incremental cost-effectiveness ratio for DFPSwith respect to cost per extra successes when it did notdominate than it had for IOPS with respect to cost permonth of symptoms avoided when it did not dominate.

If these results are to guide policy development, it isuseful to consider how the decision to choose the IOPSor DFPS option causes stress for parents. For instance,howmuchmore out-of-pocket costwould parents bewill-ing to pay to avoid additionalmonths of symptoms?Howmuch cost savings would cause them to select the less-expensive option despite the longer duration of symp-toms? Although we had to make many assumptions forthis model, repeated simulations indicated amiddle 95%range from $19 to $81 per month of symptoms avoidedunder the IOPS option. It seems that this additional cost,even if not covered by insurers, might be small enoughthat parents would consider paying the uncovered dif-ference to alleviate the symptoms sooner.

Regardless of how the results are used, a more pre-cise estimate of the spontaneous resolution rate ofNLDO is critical for establishing which approach is theclinically or economically obvious choice or whethertrade-offs between costs and outcomes need to be con-sidered carefully. The PEDIG network has launchedsuch a prospective randomized trial with the primaryobjective to more precisely estimate the rate of sponta-neous resolution. An additional primary objective is toassess the cost-effectiveness of IOPS and DFPS for chil-dren aged 6 to 9 months with symptoms from congeni-tal NLDO.

The present study is not without limitations. As inmany cost-effectiveness analyses, these limitations in-clude estimates of costs and clinical outcomes drawnfrom different sources and a model constructed tosimulate what is likely to happen. Some of the studiesfrom which data were gathered are not definitive, par-ticularly the true rate of spontaneous resolution. Thecost data are not actual costs but are Medicare-allowedreimbursements for participating providers. The Medi-care fee schedule likely represents a reasonable nationalweighting of fees for physician and facility, although ac-tual values vary under an assortment of private and gov-ernment programs. The proportion of facility cases per-formed in the HOPD compared with an ASC was drawnfrom a multicenter observational study7 for our basecase and may not represent current clinical practice, al-though we assessed the impact of site of service from0% to 100% ASC use. In addition, a number of estimatesneeded to be developed by consensus. Also, the value ofparental time in obtaining medical care for children isnot included. If parental time were included, the costper month of symptoms averted by IOPS would belower. The type of choices we made in our model is




standard in economic evaluation, but primary data col-lected during an effectiveness-oriented randomizedtrial would be superior.

In conclusion, for a 6- to 9-month-old infant withNLDO, under the assumption that the spontaneous reso-lution rate is 75%, the surgeons choice of IOPS or DFPShas little impact in terms of overall success rate or costfor a population. Because cost-effectiveness is criticallydependent on the spontaneous resolution rate, prospec-tive data are needed to refine this point estimate.

Submitted for Publication: February 12, 2010; final re-vision received August 12, 2010; accepted August 26,2010.Correspondence:Michael X. Repka, MD, MBA, c/o JaebCenter for Health Research, 15310 Amberly Dr, Ste 350,Tampa, FL 33647 ([email protected]).Financial Disclosure: None reported.Funding/Support: This study was supported in part bygrant EY011751 through a cooperative agreement withthe National Eye Institute of the National Institutes ofHealth, Department of Health and Human Services.Online-Only Material: The eTable is available at http://www.archophthalmol.com. This article is featured in theArchives Journal Club. Go to http://www.archophthalmol.com to download teaching PowerPoint slides.

REFERENCES

1. Price HW. Dacryostenosis. J Pediatr. 1947;30(3):300-305.2. Petersen RA, Robb RM. The natural course of congenital obstruction of the na-

solacrimal duct. J Pediatr Ophthalmol Strabismus. 1978;15(4):246-250.3. Paul TO, Shepherd R. Congenital nasolacrimal duct obstruction: natural history

and the timing of optimal intervention. J Pediatr Ophthalmol Strabismus. 1994;31(6):362-367.

4. Nelson LR, Calhoun JH, Menduke H. Medical management of congenital naso-lacrimal duct obstruction. Ophthalmology. 1985;92(9):1187-1190.

5. Ghuman T, Gonzales C, Mazow ML. Treatment of congenital nasolacrimal ductobstruction. Am Orthopt J. 1999;49:163-168.

6. Paul TO. Medical management of congenital nasolacrimal duct obstruction. J Pe-diatr Ophthalmol Strabismus. 1985;22(2):68-70.

7. RepkaMX, Chandler DL, Beck RW, et al; Pediatric Eye Disease Investigator Group.Primary treatment of nasolacrimal duct obstructionwith probing in children youngerthan 4 years. Ophthalmology. 2008;115(3):577-584, e3.

8. Medicare program: changes to the hospital outpatient prospective payment sys-tem and CY 2009 payment rates. In: Dept of Health and Human Services, Centersfor Medicare & Medicaid Services. Fed Regist. 2008;73(223)(addendum AA):68930-69082.

9. Medicare Program: payment policies under the physician fee. In: Dept of HealthandHuman Services, Centers forMedicare &Medicaid Services. Fed Regist. 2008;73(224):70061.

10. Repka MX, Chandler DL, Bremer DL, Collins ML, Lee DH; Pediatric Eye DiseaseInvestigator Group. Repeat probing for treatment of persistent nasolacrimal ductobstruction. J AAPOS. 2009;13(3):306-307.

11. Repka MX, Chandler DL, Holmes JM, et al; Pediatric Eye Disease InvestigatorGroup. Balloon catheter dilation and nasolacrimal duct intubation for treatmentof nasolacrimal duct obstruction after failed probing. Arch Ophthalmol. 2009;127(5):633-639.

12. Beebe M, Dalton JA, Espronceda M, et al. Current Procedural Terminology. Chi-cago, IL: American Medical Association; 2009.

In cataract operations more than in any oth-ers it is desirable that the surgeon should feelfit. If the surgeon feels ill, or tired, he shouldnot operate for cataract that day. In my ownoperating I have frequently noticed that I canoperate better on some days than on others,and that sometimes I am quite unfit to op-erate. These bad days are usually ones onwhich I have taken quinine the night be-fore, on account of having felt a little fever-ish. Until I came to recognize the bad effectit had uponme I used to operate in themorn-ing although I had taken quinine the nightbefore; the result was that I could not keepmy hand from shaking; fine tremor wouldmanifest itself whenmaking the puncture andwhen seizing the iris for iridectomy.Source:McKechnieWE. Cataract operations andthe preparation of the surgeon. Arch Ophthalmol.1911;40:20.




coste eficacia master!!!

Documents