Development and assessment of a priority score for cataract surgery

Development and assessment of a priority score for cataract surgery Maria Pia Fantini, *t MD; Antonella Negro,* CompSc; Stefano Accorsi, * CompSc; Luca Cisbani,* Biost; Francesco Taroni, * MD; Roberto Grilli,* MD ABSTRACT • RESUME Background: Point-count measures of clinical priority are increasingly put forward for managing waiting lists. However, their development does not consider explicitly the appropriateness of the indications. Furthermore, an estimate of their effect in clinical practice is needed, assessing the amount of gains and losses in terms of time waited for patients with different priority scores. Methods: We developed appropriateness criteria for cataract surgery using the RAND method and applied them to a sample of 567 patients consecutively placed on a waiting list for cataract surgery. In addition, clinicians were asked to express the priority attributed to each patient using a I0-cm visual analogue scale, where 0 = minimal priority and I0 = maximum priority. We developed a priority score, using regression analysis to identify the set of clinical characteristics that best predicted clinicians' priority rating and to estimate their individual weight. We used a computer simulation model to compare mean waiting times with management of the waiting list using the priority score and using the "first-come, first-served" approach. Results: Overall, 332 patients (60.8%) were referred for cataract surgery for indications deemed appropriate, and their mean priority rating was 5.9 (95% confidence interval [CI] 5.7-6.1 ). The corresponding figures for the 20 I (36.8%) uncertain indications and the 13 (2.4%) inappropriate indications were 4.5 (95% Cl 4.1-4.7) and 2.6 (95% Cl 1.3-3.9) respectively. The clinical characteristics included in the priority score (visual acuity in the operated eye and in the contralateral eye, visual function and ability to live or work independently) accounted for 35% of the variance in clinicians' ratings of priority. In the computer simulation model, patients with the highest priority experienced a variable reduction in mean waiting time (9% to 27%) depending to how time spent waiting was integrated into the clinical score. Interpretation: We conclude that the use of priority ratings in the management of a waiting list for cataract surgery leads to results that maintain the desirable coherence between priority and appropriateness of indication. The results also suggest that the implementation in clinical practice of priority scores may be worth the effort, given the potential reduction in waiting time for patients at high priority.

From *the Agenzia Sanitaria Regionale Emilia-Romagna, Bologna, Italy, and tthe Istituto di Igiene, Universita di Bologna, Bologna, Italy

Department of Clinical Governance, Viale Aldo Moro 21, Bologna, Italy; fax +39-051-6397053, rgrilli@ asr.regione.emilia-romagna.it

Originally received Feb. 13, 2003 Accepted for publication Sept. 25, 2003

This article has been peer-reviewed.

Correspondence to: Dr. Roberto Grilli, Agenzia Sanitaria Regionale,

Can j Ophthalmol 2004;39:48-55

48

Priority score for cataract surgery-Fantini et al

Priority score for cataract surgery-Fantini et al Contexte : On met de plus en plus de l'avant le recours a des systemes de pointage pour gerer les listes d'attente. Le dressage de ces listes ne tient cependant pas compte explicitement du degre de pertinence des indications. L'on a davantage besoin d'une estimation approximative des consequences que pourrait avoir sur le plan clinique de tels systemes. Methodes : Nous avons elabore des criteres de pertinence pour !'operation de Ia cataracte selon Ia methode RAND, que nous avons appliques a un echantillonnage de 567 patients inscrits de fac;:on consecutive sur une liste d'attente pour cette chirurgie. En outre, les cliniciens ont ete invites a exprimer l'ordre de priorite attribue a chaque patient selon une echelle visuelle analogue de I0 em, ou 0 = Ia priorite minimale et I0, Ia maximal e. Nous avons elabore un ordre de priorite a partir d'une analyse de regression pour etablir le jeu des caracteristiques cliniques permettant de prevoir au mieux Ia cote de priorite des cliniciens et d'estimer leur poids individuel. Nous avons utilise des modeles de simulation informatique pour comparer Ia moyenne des delais d'attente avec Ia gestion des listes selon le pointage de priorite et en regard de !'approche « premier arrive, premier servi ». Resultats: Dans I' ensemble, on a juge que !'indication de Ia chirurgie de Ia cataracte avait ete pertinente pour 332 patients (60,8 %), dont Ia cote moyenne de priorite etait de 5,9 (intervalle de confiance [IC] a95 % 5,7-6,1 ). Les cotes correspondantes pour les 20 I indications jugees incertaines (36,8 %) et les 13 inopportunes (2,4 %) etaient respectivement de 4,5 (IC a 95% 4,1-4,7) et de 2,6 (IC a 95% 1,3-3,9). Les caracteristiques cliniques des cotes de priorite (acuite visuelle dans l'reil opere et l'reil oppose, fonction visuelle et capacite de vivre ou de travailler de fac;:on autonome) representaient 35 %de Ia variance chez les cotes de priorite des cliniciens. Dans le modele de simulation informatisee, les patients qui avaient Ia plus haute cote de priorite ont vu baisser de fac;:on variable Ia moyenne du delai d'attente (de 9 % a 27 %) selon le mode d'integration du delai a Ia cote clinique. Interpretation : Nous en concluons que le recours a des systemes de pointage pour gerer les listes d'attente pour Ia chirurgie de Ia cataracte entraine des resultats qui maintiennent Ia coherence souhaitable entre l'ordre de priorite et Ia pertinence de !'indication. Les resultats indiquent aussi que !'application des cotes de priorite en milieu clinique peut valoir Ia peine, vu Ia possibilite de reduire le delai d'attente des patients pour lesquels Ia cote de priorite est elevee.

aiting lists for elective care are a complex political, organizational and clinical problem faced by most publicly funded national health care systems. As reliance on increasing supply alone has been shown to have a limited effect in the long run, 1 a great deal of attention is being paid to clinical tools, such as practice guidelines, explicit referral criteria and clinical priority scores, for making access to elective procedures more selective. The assumption underlying these approaches is that the number of patients waiting could be reduced by ensuring that only those referred for appropriate clinical indications are placed on the list and that waiting times for those on the list reflect individual patients' needs. In addition, man-

W

agement of waiting lists would be more transparent and based on shared criteria. The development of point-count priority criteria for a variety of clinical conditions has attracted considerable attention. 2- 5 In general, these instruments are based on the attribution to each individual patient of a score, based on the weight (i.e., relevance) assigned to each clinical or social factor. However, clinical priority criteria are not problemfree. Methods for their development are to a great extent based on consensus among experts, who make explicit the clinical factors deemed to be relevant in determining priority. Therefore, one could question the reliability of the process (i.e., whether different

CAN J OPHTHALMOL-VOL. 39, NO. I, 2004

49

Priority score for cataract surgery-Fantini et al

groups of experts using the same method identify the same set of clinical characteristics as relevant) and its ability to keep priority in line with the distinct, although closely related, dimension of appropriateness (i.e., avoiding assignment of a high priority score to patients with inappropriate clinical indications). Furthermore, the effect of these scoring systems when implemented in the management of waiting lists has not been fully explored. In particular, it would be relevant to know the potential gain (in terms of time spent waiting) for patients assigned to various priority categories as compared with the current dominant "first-come, first-served" approach. Recently, within the framework of federalist reform of Italy's entire institutional system, regional health authorities were given increasing autonomy in the management of health care services and full responsibility for reducing waiting times for elective care. Some regional health authorities chose to increase the level of supply, whereas others, including Emilia-Romagna, opted to focus mainly on demand, through the promotion in clinical practice of tools and instruments enabling health care services to discriminate between appropriate and inappropriate referrals. In this context, we developed priority criteria for cataract surgery using a method similar to that recently adopted in the Western Canada Waiting List Project. 5·6 In this paper we report the results, focusing on the relation between priority and appropriateness for this surgical intervention and on assessment of the effect of the priority system on waiting times through a computer simulation model.

In our context the multidisciplinary expert panel was convened at the end of 1999 and included five university and community ophthalmologists, two family physicians, an optometrist, a health services researcher and a health information specialist. The panel was chaired by one of us (M.P.F.). The panel met three times, and the final results were available in December 2000. The factors considered in the development of the clinical scenarios included type of surgery (unilateral or bilateral, with or without coexisting disease), visual acuity (in the operated eye and in the contralateral eye), visual function, life expectancy and patient's ability to work or live independently. Assessment of appropriateness and priority

We collected information on the clinical characteristics of a consecutive sample of patients referred for cataract surgery during June 2000 at five centres (two university and three general hospitals) using a standardized data collection form. In addition, clinicians were asked to express the priority attributed to each patient using a 10-cm visual analogue scale, where 0 = minimal priority and 10 = maximum priority. The appropriateness criteria developed by the expert panel were applied to the sample, and rates of appropriate, inappropriate and uncertain indications were estimated. We then compared mean priority ratings for the three types of indication using analysis of variance (ANOVA).

METHODS

Development of appropriateness criteria We developed appropriateness criteria for cataract surgery using the RAND method/·8 a structured consensus process based on a modified Delphi technique that has been used extensively in health care for assessing the appropriateness of procedures and interventions. A list of clinical scenarios, each representing a hypothetical patient, is presented to a multidisciplinary panel of experts, who are asked to individually rate the appropriateness of use of a specific intervention on a scale of 1 to 9, where a score of 1-3 indicates inappropriateness, 7-9 appropriateness, and 4-6 a judgement of uncertainty. The clinical scenarios are drawn from the patient characteristics affecting the decision-making process concerning the use of the intervention in question.

50

CAN JOPHTHALMOL-VOL. 39, NO. I, 2004

Identification of patient characteristics related to degree of priority assigned, and development of priority score The relation between individual priority ratings drawn from the visual analogue scale and patient clinical characteristics was explored both visually, through graphic displays, and statistically, by comparing mean priority values across different clinical characteristics using ANOVA. After these univariate analyses, factors with a statistically significant association were entered into a multivariate linear regression analysis, with the priority rating as the dependent variable. Standardized regression coefficients representing the weight attributed to each factor were estimated and then translated to obtain a composite clinical score ranging from 0 to

Priority score for cataract surgery-Fantini et al

100. We measured the extent of correlation among criteria using standard Pearson r statistics. Assessment of effect of priority scoring system on waiting times We designed a simulation model that allowed comparison between a hypothetical list managed on the basis of the first-come, first-served approach and another hypothetical list with similar characteristics (in terms of patient number and case mix) but with access managed through the use of the priority score. The model included a weight assigned to time spent waiting by each patient, in order to avoid a situation in which a patient with relatively low priority on clinical grounds never gained access to the procedure as other patients with higher priority scores were placed on the list. We modelled time spent waiting in two different ways, according to a linear function and according to a cubic function. In the first case, patients receive, in addition to the priority score, an equal point for each day spent waiting. In the second case the amount added to the clinical score varies over time according to a cubic function. Therefore, the overall score (OS) for each patient was OS = T + S with the linear model, and OS = (T 3/U) + S with the cubic model, where T = time spent waiting and S = priority score, standardized in such a way that it ranges from 0 to L, with L =mean length of time (in days) on the list. The simulation was based on the following assumptions: • the average wait was 180 days when the list was managed with the first-come, first-served approach; • supply and demand varied according to a Poisson distribution of parameters 'A, where ')..in and ')..out were the regional number of patients placed on the waiting list (n = 120 per day) and the average daily number of interventions performed (n = 106) respectively (these parameters varied in sensitivity analyses); • the dynamic of the hypothetical list was observed over a timeframe of 1000 days; and • the distribution of patients within the various classes of priority score was drawn from that observed in the sample of patients recruited into the study.

Table !-Demographic and clinical characteristics of 567 patients referred for cataract surgery Characteristic Sex Female Male Visual acuity in operated eye ~ 8/10 5/10-7/10 2/10-4/10 < 2/10 Visual acuity in contralateral eye Blind ~ 5/10 2/10-4/10 < 2/10 Visual function No impairment Mild to moderate impairment Severe impairment Ability to live/work independently Not threatened or no difficulties Not threatened but difficult Threatened but not immediately Immediately threatened or unable Type of cataract Unilateral without coexisting disease Unilateral with coexisting disease Bilateral without coexisting disease Bilateral with coexisting disease Coexisting ocular disease

No. (and%) of patients* 361 (63.7) 206 (36.3) 8 107 217 225

( 1.4) (19.2) (39.0) (40.4)

10 374 116 59

(1.8) (66.9) (20.8) (10.6)

34 (6.1) 330 (59.4) 191 (34.4) 119 254 108 73

(21.5) (45.8) (19.5) ( 13.2)

152 91 191 132 184

(26.8) (16.1) (33.7) (23.3) (32.4)

*Subtotals may vary owing to missing data.

RESULTS

A total of 567 patients (including 225 from the two university centres) were enrolled over the study period. The patients ranged in age from 33 to 100 years (mean 75 [standard deviation 9] years, median 77 years). The characteristics of the sample are shown in Table 1. According to the expert panel criteria, 332 (60.8%) of the indications were judged as appropriate, 201 (36.8%) as uncertain and 13 (2.4%) as inappropriate. The mean priority rating assigned by clinicians to individual patients on the visual analogue scale was 5.3 (median 6, range 1-10). Priority ratings were related to the appropriateness of the indication, with

CAN J OPHTHALMOL-VOL. 39, NO. I, 2004

51

Priority score for cataract surgery-Fantini et al

Table 2-Mean priority rating according to the appropriateness of the clinical indication for cataract surgery Appropriateness of indication

No. (and %) of cases*

Appropriate Uncertain Inappropriate Total

332 (60.8) 201 (36.8) 13 (2.4) 546 (100.0)

Mean rating (and 95% confidence interval)t 5.9 4.5 2.6 5.3

(5.7--6.1) (4.1-4.7) ( 1.3-3.9) ( 1.0-1 0.0)

*Twenty-one cases were excluded owing to missing information. tOne-way analysis of variance F test 43.821, p < 000.1.

Table 3-Coefficients from linear regression analysis of the association between patient characteristics and priority rating Variable Life expectancy Visual acuity in contralateral eye Visual function Visual acuity in operated eye Ability to live/work independently Coexisting disease Age

Coefficient

p value

0.029

0.490

0.162 0.204 0.203

0.000 0.000 0.000

0.306 0.100 -0.001

0.000 0.017 0.985

patients with appropriate indications receiving higher priority ratings than those whose indications were uncertain or inappropriate (Table 2).

Characteristics related to priority rating assigned and development of priority score Results from the multiple linear regression analysis aimed at exploring the relation between patient characteristics and priority ratings are shown in Table 3. Covariates associated with priority ratings were visual acuity in the operated eye and in the contralateral eye, visual function and ability to live or work independently. The R2 with this set of criteria was 36% (adjusted R2 35%), indicating that the criteria could explain more than one-third of the statistical variance in clinicians' priority ratings. In general, intercriteria correlation was quite low, with the Pearson correlation coefficient never greater than 0.41 (data not shown), indicating that contributions of the criteria regressed on visual analogue scale priority ratings were relatively independent.

52

CAN J OPHTHALMOL-VOL. 39, NO. I, 2004

Table 4-Weight assigned to the selected variables included in the clinical priority score Variable Visual acuity in operated eye ~ 8/10 5/10-7/10 2/10-4/10 < 2/10 Visual acuity in contralateral eye Blind ~ 5/10 2/10-4/10 < 2/10 Visual function No impairment Mild to moderate impairment Severe impairment Ability to live/work independently Not threatened or no difficulties Not threatened but difficult Threatened but not immediately Immediately threatened or unable

Level

Weight

I 2 3 4

0 II 17 23

I 2 3 4

0 7 10 13

2 3

0 14 21

I 2

0 22

3

32

4

43

Table 4 shows the weight assigned to each level of the covariates associated with priority ratings.

Effect of priority score on waiting time Patients with the highest priority score (91-100) had a mean wait of 197.4 days when the first-come, firstserved approach was used. With management of the waiting list according to clinical priority score, the mean

Priority score for cataract surgery-Fantini et al

Table 5-Comparison of time waited with management of the waiting list according to the clinical priority score and with a "first-come, first-served" approach Management with priority score Time modelled according to linear function

Time modelled according to cubic function

Management with first-come, first-served approach

Overall priority score

Minimum wait, d

Maximum wait, d

Mean wait, d

Minimum wait, d

Maximum wait,d

Mean wait, d

Minimum wait,d

Maximum wait, d

Mean wait, d

100-91 90-81 80-71 70-61 60-51 50-41 40-31 30-21 20-11 10-1

122 140 157 175 192 210 227 248 272 298

180 181 201 221 240 260 277 299 308 313

144.5 160.6 177.7 193.9 212.3 234.8 251 .8 273.6 290.4 303.6

158 166 172 178 184 189 195 201 207 213

199 206 211 217 223 228 232 237 239 240

179.2 185.0 191.4 197.0 202.4 209.2 213.7 219.5 222.9 220.0

179 179 179 179 179 179 179 180 180 182

215 215 215 214 214 215 214 210 189 209

197.4 197.6 197.8 197.4 197.8 197.9 197.6 198.0 197.7 193.3

wait was 144.5 days with modelling of time according to a linear function, corresponding to a relative reduction in average time waited of 27% (Table 5). The corresponding figure when time was modelled according to a cubic function was 179.2 days (relative reduction 9%). With time modelled according to a linear function, patients with the lowest priority score (1-10) had a mean wait of 303.6 days, compared with 193.3 days with the first-come, first-served approach, corresponding to an increase in waiting time of 57% (Table 5). The pattern of results when time was modelled as a cubic function was similar, although differences in mean waiting time across priority classes were reduced. INTERPRETATION

Although our study included a relatively limited sample, and despite the low prevalence (2%) of clinical indications deemed to be inappropriate, our findings are reassuring, at least to the extent to which they show that appropriateness and priority ratings appear to be correlated in the expected and desirable direction, with patients referred for appropriate indications getting higher ratings than those referred for uncertain or inappropriate indications. In developing the priority score we used a method similar to that adopted in the Western Canada Waiting List Project. 5 It is worth noting that the subjective priority ratings assigned by clinicians in our study led

to the identification of a set of factors (visual acuity, visual function and ability to live or work independently) similar to those identified in the Canadian project. The set of variables identified was able to explain only part of the overall variance in priority ratings (35% in our study and 32% in the Western Canada Waiting List Project). Nevertheless, these variables proved to have face validity and credibility when presented to the ophthalmologists participating in our study, as they did in the Canadian experience. We attempted to assess the potential effect on waiting time of using clinical priority scores in managing waiting lists. The results of our simulation model indicate that patients with the highest priority scores may experience a substantial reduction in waiting time with the adoption of these criteria. These findings emerged consistently across our sensitivity analyses, although affected by the way in which time already spent on the waiting list was integrated into the clinical score. Adding time to the clinical score according to a linear function maximized the benefits (in terms of days potentially gained) for patients with the highest priority scores as well as the losses for those with lower scores. With time modelled according to a cubic function, gains could be so negligible as to bring into question the value of implementing a priority system for managing waiting lists. The relevance of waiting times can be considered from a clinical as well as from a policy perspective.

CAN J OPHTHALMOL-VOL. 39, NO. I, 2004

53

Priority score for cataract surgery-Fantini et al

~~ -

100

E :;" c. 0

1!-

80 60

..Y

40 20 0

.

.....--'

~ 0

"

20

26

34

40

44

.-·

High priority rating(;;:: 6)

- - Low priority rating (< 6)

---·

"~ .

~

I

~

-------

49

,

62

67

75

85

91

110116127133 148154193

Time waited, d

Fig. 1-Actual time to procedure for 230 patients referred for cataract surgery whose priority was assessed by ophthalmologists on a visual analogue scale of 0 to I0.

Whereas in clinical terms the relevance of waiting is related to the extent to which a timely diagnosis or treatment may affect the patient's prognosis (and therefore may be variable according to the characteristics of the disease or clinical condition9•10), from a policy perspective what is really important is how patients perceive the time they have to wait. In this light, some technical issues, such as how time could be better handled in the context of a priority score, are related to the policy problem of identifying a trade-off between gains and losses that is reasonable and acceptable to patients and society. As the implementation of a priority score may be demanding in organizational terms, its effect on waiting times must be worth the effort. In particular, as some form of prioritization is already used by clinicians in current clinical practice (although based on implicit and variable criteria), it is important to check whether the system of waiting list management already in place actually succeeds in assuring quicker access to elective procedures for those considered at higher priority. We have been able to address this aspect using data for 230 patients (40% of our original sample) for whom information on actual waiting time was available. Fig. 1 shows actual waiting times for patients classified according to the priority ratings assigned by clinicians. Those considered at highest priority (i.e., with a score higher than 6 on the visual analogue scale) experienced longer waits. Thus, in this case, the current management of waiting lists does not achieve the goal of assuring rapid access to those with greatest clinical need. Our study has several limitations. First, the sample size was relatively limited; however, our findings are in keeping with those of other studies. 6 In addition, our assessment of the effect of the use of the priority

54

CAN J OPHTHALMOL-VOL. 39, NO. I, 2004

score on waiting times was based on a computer simulation model rather than on empirical evaluation of its effect on clinical practice. However, simulation models are widely used to explore ways of managing waiting lists 11 - 16 and are deemed to be a valuable preliminary step before undertaking the major effort of implementing in clinical practice systems that are expected to be complex and demanding. In conclusion, our results confirm the applicability and reliability of the approach developed in the Western Canada Waiting List Project. 5•6 We thus conclude that this approach leads to results that maintain the desirable coherence between priority and appropriateness of indication. In addition, our results suggest that the implementation in clinical practice of priority scores is worth the effort, given the potential reduction in waiting time for patients deemed to be at high priority. The following clinicians and managers contributed to the development of appropriateness and priority criteria and to data collection: Candia Calanchi, Domenico Cucinotta, Fausto Marchetta and Martina Taglioni (Azienda Ospedaliera S.Orsola-Malpighi, Bologna), Franco Chiaravallotti and Luca Sircana (Azienda Ospedaliera Reggio-Emilia), Giovanna Costantini (Associazione ltaliana Ospedalita Privata), Giovanni Maraini and Cristina Bonetti (Azienda Ospedaliera Parma) and Giovanni Pirazzoli and Maria Grazia Stagni (Azienda USL Cesena). REFERENCES

1. Harrison A, New B. Access to elective care. What should really be done about waiting lists? London: King's Fund Publishing; 2000. 2. BeHan L, Mathen M. The Manitoba Cataract Waiting List Program. CMAJ 2001;164:1177-80. 3. Radom DC. Setting priorities for waiting lists: defining our terms. Steering Committee of the Western Canada Waiting List Project. CMAJ 2000;163:857-60. 4. Radom DC, Holmes AC. The New Zealand priority criteria project. Part 1: Overview. BMJ 1997;314:131-8. 5. Noseworthy TW, chairperson. From chaos to order: making sense of waiting lists in Canada. Final report of the Western Canada Waiting List Project. Calgary: Western Canada Waiting List Project; 2001. Available: www. wcw 1. org/media/pdf/library /final_reports. 2. pdf (accessed 2003 Dec 3). 6. Romanchuk KG, Sanmugasunderam S, Radom DC. Developing cataract surgery priority criteria: results from the Western Canada Waiting List Project. Steering Committee of the Western Canada Waiting List Project. Can J Ophthalmol2002;37:145-54. 7. Brook R, Chassin M, Park R. A method for detailed assessment of the appropriateness of medical technolo·

Priority score for cataract surgery-Fantini et al

gies. Inti J Techno! Assess Health Care 1986;2:53-63. 8. Fitch K, Bernstein SJ, Aguilar MD, Burnand B, LaCalle JR, Lazaro P, et al. The RAND/UCLA appropriateness method user's manual. Santa Monica (CA): RAND Corporation; 2001. 9. Hurst NP, Lambert CM, Forbes J, Lochhead A, Major K, Lock P. Does waiting matter? A randomized controlled trial of new non-urgent rheumatology out-patient referrals. Rheumatology 2000;39:369-76. 10. Hajat S, Fitzpatrick R, Morris R, Reeves B, Rigge M, Williams 0, et al. Does waiting for total hip replacement matter? Prospective cohort study. J Health Serv Res Policy 2002;7:19-25. 11. Aharonson-Daniel L, Paul RJ, Hedley AJ. Management of queues in out-patient departments: the use of computer simulation. J Manag Med 1996;10:50-8, 3. 12. Burkell J, Wright A, Hoffmaster B, Webb K. A decisionmaking aid for long-term care waiting list policies: mod-

13. 14.

15.

16.

elling first -come, first -served vs. needs-based criteria. Healthc Manage Forum 1996;9:35-9. Huarng F, Lee MR. Using simulation in out-patient queues: a c:ase study. Int J Health Care Qual Assur 1996;9:21-5. Clague JE, Reed PG, Barlow J, Rada R, Clarke M, Edwards RH. Improving outpatient clinic efficiency using computer simulation. Int J Health Care Qual Assur Inc Leadersh Health Serv 1997;10:197-201. Ratcliffe J, Young T, Buxton M, Eldabi T, Paul R, Burroughs A, et al. A simulation modelling approach to evaluating alternative policies for the management of the waiting list for liver transplantation. Health Care Manag Sci 2001;4:117-24. Tuft S, Gallivan S. Computer modelling of a cataract waiting list. Br J Ophthalmol2001;85:582-5.

Key words: cataract extraction, waiting lists, priority criteria, appropriateness

CAN J OPHTHALMOL-VOL.39, NO. I, 2004

55