Article Text

Monitoring ocular hypertension, how much and how often? A cost-effectiveness perspective
  1. R Hernández1,
  2. JM Burr2,
  3. L Vale3,
  4. A Azuara-Blanco4,
  5. JA Cook5,
  6. K Banister6,
  7. A Tuulonen7,
  8. M Ryan1
  9. for the Surveillance of Ocular Hypertension Study group
    1. 1Health Economics Research Unit, Institute of Applied Health Sciences, University of Aberdeen, Aberdeen, UK
    2. 2School of Medicine, University of St Andrews, St Andrews, Fife, UK
    3. 3Health Economics Group, Institute of Health & Society, Newcastle University, Newcastle upon Tyne, UK
    4. 4School of Medicine, Dentistry and Biomedical Sciences, Queen's University Belfast, Belfast, UK
    5. 5Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, Centre for Statistics in Medicine, University of Oxford, Botnar Research Centre, Nuffield Orthopaedic Centre, Oxford, UK
    6. 6Health Services Research Unit, Institute of Applied Health Sciences, University of Aberdeen, Aberdeen, UK
    7. 7Tays Eye Centre, Tampere University Hospital, Tampere, Finland
    1. Correspondence to Dr J M Burr, School of Medicine, University of St Andrews, St Andrews, Fife KY16 9TF, UK; jmb28{at}st-andrews.ac.uk

    Abstract

    Objective To assess the efficiency of alternative monitoring services for people with ocular hypertension (OHT), a glaucoma risk factor.

    Design Discrete event simulation model comparing five alternative care pathways: treatment at OHT diagnosis with minimal monitoring; biennial monitoring (primary and secondary care) with treatment if baseline predicted 5-year glaucoma risk is ≥6%; monitoring and treatment aligned to National Institute for Health and Care Excellence (NICE) glaucoma guidance (conservative and intensive).

    Setting UK health services perspective.

    Participants Simulated cohort of 10 000 adults with OHT (mean intraocular pressure (IOP) 24.9 mm Hg (SD 2.4).

    Main outcome measures Costs, glaucoma detected, quality-adjusted life years (QALYs).

    Results Treating at diagnosis was the least costly and least effective in avoiding glaucoma and progression. Intensive monitoring following NICE guidance was the most costly and effective. However, considering a wider cost–utility perspective, biennial monitoring was less costly and provided more QALYs than NICE pathways, but was unlikely to be cost-effective compared with treating at diagnosis (£86 717 per additional QALY gained). The findings were robust to risk thresholds for initiating monitoring but were sensitive to treatment threshold, National Health Service costs and treatment adherence.

    Conclusions For confirmed OHT, glaucoma monitoring more frequently than every 2 years is unlikely to be efficient. Primary treatment and minimal monitoring (assessing treatment responsiveness (IOP)) could be considered; however, further data to refine glaucoma risk prediction models and value patient preferences for treatment are needed. Consideration to innovative and affordable service redesign focused on treatment responsiveness rather than more glaucoma testing is recommended.

    • Intraocular pressure
    • Glaucoma
    • Public health
    View Full Text

    Statistics from Altmetric.com

    Introduction

    Avoiding sight loss is a public health priority1 but decisions have to be made in terms of how best to manage eye care across competing demands. In the UK, about 24 000 people are newly registered with sight loss each year (blind and partial sight) with glaucoma being second to macular degeneration as the leading cause.2 ,3 Although ocular hypertension (OHT) is the main and only modifiable risk factor for glaucoma,4 ,5 organising a monitoring programme to monitor intraocular pressure (IOP) and detect early glaucoma has the potential to overburden healthcare and patients. Choices have to be made.

    The UK National Institute for Health and Care Excellence (NICE) clinical guideline for glaucoma6 recommends long-term monitoring of OHT in specialist (healthcare professional accredited in glaucoma)-led service either in secondary care (consultant-led hospital eye service) or primary care (community optometry) depending on local commissioning arrangements. Thresholds for initiating ocular hypotensive treatment are defined proportionate to glaucoma risk with stratification based upon the findings of a Markov economic model.6 Recommended monitoring intervals were informed by a budget impact analysis and expert opinion. However, concerns were subsequently raised that glaucoma services are overburdened with monitoring low-risk disease.7

    This study compares plausible alternative monitoring programmes for OHT (varying monitoring intervals, setting and treatment thresholds) to inform eye-care policy in the UK. The study is part of a wider programme of work commissioned by the National Institute of Health Research, Health Technology Assessment programme to determine the optimum surveillance for people with OHT and is reported elsewhere in full.8

    Methods

    We developed a discrete event simulation model8 to assess the relative efficiency of monitoring a simulated cohort of 10 000 people with confirmed OHT (defined as an IOP >21 mm Hg and no clinical signs of glaucoma) in the UK. We estimated cost-effectiveness (cost per glaucoma cases detected) and cost–utility (cost per incremental quality-adjusted life year (QALY) gained) over 20 years, adopting a NHS perspective and discounted costs and benefits at the recommended 3.5% discount rate.9 All costs are reported in 2009–2010 Pound Sterling. The model structure is described in full elsewhere.8

    In brief, we compared five care pathways: three ‘new’ care pathways and two based upon the existing NICE recommended pathways for OHT, that is, guidance for current care, see table 1. The three ‘new’ care pathways were (i) monitoring people with OHT biennially in secondary (consultant-led eye care) or (ii) biennially in primary eye care (glaucoma-trained community clinical health professional (optometrist or General Practitioner)-led care). In both ‘biennial’ pathways, ocular hypotensive treatment was modelled when the baseline risk of glaucoma was ≥6% based on a glaucoma 5-year risk estimator.10 For those requiring treatment, responsiveness was assessed at 2 months and a <15% reduction in IOP prompted treatment escalation by adding a topical beta-blocker. As an extreme alternative, we also modelled a pathway where for confirmed OHT (iii) treatment was initiated irrespective of glaucoma risk stratification with annual IOP monitoring by community optometrist. A <15% reduction in IOP prompted referral into secondary care with monitoring according to NICE OHT pathway.6 The NICE monitoring guidance was summarised as (iv) an intensive pathway representing a monitoring interval between 4 and 12 months and (v) a more conservative pathway with monitoring intervals between 6 and 24 months. These two NICE-informed pathways reflect variations in the recommended monitoring frequency in the NICE guideline.6 The monitoring interval in both NICE pathways depended on baseline risk stratification based on age, IOP and central corneal thickness (CCT), factors regarded as predictors of glaucoma (see online supplementary appendix table A1 for a detailed description of the NICE-informed pathways included in our model). In all five pathways, if conversion to glaucoma occurred, subsequent care was in the hospital eye service according to the NICE glaucoma treatment guideline.6

    Table 1

    Modelled care pathways for monitoring confirmed OHT

    Apart from the ‘treat all’ pathway, each simulated individual within the model had a predefined glaucoma risk, which was based on the natural history of open-angle glaucoma estimated from the distribution of predictors in a European-derived population.11 The mean age of the simulated population cohort was 57 years, with lower and upper interquartile range limits of 51 and 64, respectively, and mean IOP of 25 mm Hg ranging from 21 to 36 mm Hg. The underlying distribution of IOP in the simulated cohort over time was estimated from predictions of a mixed linear regression model based on IOP data from an untreated OHT cohort and described in detail elsewhere.8 The uncertainty due to measurement variability according to tonometer used was also included.12

    We populated the model with data from systematic reviews and statistical modelling estimates of the variability in IOP measurements and visual field indices8 based on data from the placebo arms of two randomised controlled OHT treatment trials.11 ,13 Details are described in table 2.

    Table 2

    Model data (natural history, efficacy and diagnostic accuracy/agreement)

    Data on the adherence to treatment (eye drops) were sparse and varied widely. We assumed that, based on expert opinion (three ophthalmologists), treatment adherence was 75% except for the ‘treat all’ monitoring pathways where we assumed lower adherence, 50%. Non-adherence was modelled as if untreated. Cost and utility data, see table 3, were derived from the British National Formulary and data from 255 people with glaucoma18 valued using UK population tariffs. We assumed that the score for those with OHT was the same as for mild glaucoma.

    Table 3

    Costs and utility estimates

    We addressed uncertainty by running one-way sensitivity analyses on a simulated cohort of 1000 individuals. Sensitivity analyses included changing the following parameters in the biennial pathways: (i) increasing the glaucoma risk threshold for initiating surveillance (eg, from 6% to 50% as opposed to surveillance for all with confirmed OHT as in the base case); (ii) increasing the glaucoma risk threshold for initiating treatment (from ≥6% in the base case to a 50% 5-year predicted risk of glaucoma) to explore the effects of treating only a high-risk group and (iii) changes to the unit price of treatment with a prostaglandin analogue (a. pricing at 50% of the value in table 3 and b. pricing prostaglandin analogue at beta blocker prices—table 3); (iv) varying the price of monitoring visits (exploring the upper and lower limits of the IQR in the NHS reference costs) and (v) varying adherence from 50% to 20% for the ‘Treat all’ pathway and 75% to 95% for the more active monitoring pathways.

    Results

    The baseline characteristics of the simulated individual characteristics are described in table 4.

    Table 4

    Simulated individual characteristics at start of the model, n=10 000

    Over a 20-year time horizon, the risk of converting to early glaucoma using the five monitoring strategies varied between 21% and 23%. The ‘treat all’ pathway (ocular hypotensive drops initiated irrespective of glaucoma risk, annual IOP monitoring in primary care with referral to secondary care if IOP treatment response was inadequate) was the least costly and least effective in terms of number of people progressing to glaucoma, table 4. NICE intensive monitoring was the most effective in avoiding conversion to and progression of glaucoma but the most costly pathway. Taking a broader perspective, by including quality of life, the differences in QALYs between any of the monitoring pathways are small as only a modest number of people in the cohort progress to any stage of glaucoma. Biennial monitoring in secondary care is more costly but provides more QALYs than a ‘treat all’ pathway. However, the incremental cost-effectiveness ratio (ICER), in moving from the ‘treat all’ pathway to biennial monitoring in secondary care, of around £87 000 is much larger than the usual maximum threshold of willingness to pay for a QALY adopted by NICE in considering implementation of new policies (£20–£30 000 per QALY gained).9

    A summary of the findings of the sensitivity analyses is presented as online supplementary appendix 2. The results were not sensitive to varying the risk threshold for initiating surveillance (6%–50%) but were sensitive to increasing the threshold for treatment. When the treatment threshold was increased to a 5-year glaucoma risk of 15% the biennial monitoring pathway (secondary care) dominated the ‘treat-all’ pathway (less costly and resulted in more QALYs). Sensitivity analyses around unit costs of treatment did not change the ICERs substantially. Varying the unit price of treatment with a prostaglandin analogue reduced the average costs for all pathways particularly those that involved more treatment (‘treat all’), but as expected no effect was observed on QALYs as only the unit costs of treatment changed.

    The findings of the base case analysis were sensitive to varying unit costs for healthcare visits for a glaucoma-monitoring visit (including perimetry) and visits to measure IOP in response to treatment. If the unit cost of hospital visit was reduced from the base case value (£180) to £73 and an IOP only visit from £90 to £51, biennial monitoring in secondary care was cost-effective compared with a ‘treat-all’ pathway, ICER £10 857 per QALY gained. Retaining the hospital glaucoma assessment at base case level (£180) but reducing the costs of an IOP visit, if the unit cost of an eye-care visit was £50–£55, biennial monitoring in secondary care was cost-effective, ICER £11 410 per QALY gained.

    The results were also sensitive to varying the estimated adherence to ocular hypotensive treatment. When adherence decreased for a ‘treat-all ‘pathway, the cost-effectiveness of biennial monitoring improved; for example, assuming a 30% adherence in the ‘treat all’ pathway and 75% adherence in a biennial pathway the ICER is £26 334.

    Discussion

    We modelled less-intensive pathways than recommended in the current NICE glaucoma guideline6 to explore optimal treatment thresholds and monitoring intervals for people with OHT.

    Additionally, we used findings from statistical models of cohort data to inform an individual's risk stratification and the frequency of monitoring in our biennial monitoring pathways. These data were not available when the NICE clinical guideline was developed. Although effective at reducing the incidence of glaucoma, based on this study's findings, pathways based on current NICE-recommended pathways are unlikely to be cost-effective from a UK policy perspective compared with the alternative approaches we explored. Initiating treatment as soon as OHT is identified with minimal monitoring once the target IOP is reached is the least costly approach, and compared with the alternative pathways modelled could be seen as the most cost-effective (balancing cost and QALYs gained). Alternatively, based on the sensitivity analyses, initiating treatment if the 5-year glaucoma risk was >10%, minimising the cost of repeat eye-care visits and supporting patient adherence to treatment, with subsequent glaucoma tests every 2 years could be an efficient approach for surveillance of OHT. There are uncertainties regarding the optimal glaucoma risk threshold for initiating treatment, service costs and long-term adherence to medication. The impact of any treatment side effects may not be fully captured in QALY estimates.

    These findings should be interpreted bearing in mind the limitations of the data available to populate the model. In particular, natural history data represented ocular hypertensive populations included in randomised controlled trials and are not necessarily generalisable to those with OHT presenting to routine eye-care services.

    We used a discrete event simulation model, which simulates the costs and consequences of individual patients, enabling disease modelling and event complexities that are difficult to model in simpler more commonly used modelling approaches such as decision tree or Markov models. In any modelling exercise there are uncertainties in model structure, that is, are the care pathways modelled appropriate, parameter uncertainty regarding the data used and time horizon of the model? We used recommended best practice methods to ensure validity of the model. This included extensive validation and calibration exercises. As part of the model implementation process, we verified, validated (to verify the model fits the empirical data) and calibrated the model where needed. We simulated individual characteristics for a population of 1000 people at the point where they enter into the model. The data were interpreted alongside data from the literature to judge face validity.11 In addition, we ran the model for shorter time horizons and confirmed that glaucoma conversion was consistent with the literature.10 ,19

    We developed the care pathways using the NICE glaucoma guideline as a representation of current recommended practice for managing OHT. This may not reflect actual practice around the UK but was a standardised comparator. We developed two pathways capturing the extremes of the NICE-recommended pathways, and based on these extremes the cost-effectiveness of anything in between could be inferred. As there is limited capacity in hospital eye-care services, we developed alternative pathways to test whether less-intensive monitoring would be a cost-effective option for a national health service. These alternative monitoring scenarios were developed in consultation with ophthalmologists, nurses, patients and the public8 and a recent economic evaluation in the Netherlands that suggested that a treating OHT irrespective of glaucoma risk was a cost-effective approach in a Dutch context.20 As neither of the extreme scenarios based on NICE were cost-effective compared with minimal glaucoma monitoring, it is unlikely that any other strategy consistent with NICE guidance would be cost-effective. However, it should be noted that the original version of the model did not allow for people deemed at low risk of developing glaucoma (CCT >590 µm and IOP <32 mm Hg) to be discharged after having a stable IOP for more than 3 years. We added this condition in a subsequent model development and the base case results reported in table 5 remain robust. Discharged individuals move to yearly check-ups at a community optometrist and while the NICE-informed strategies become, on average, less costly, they are not cost-effective compared with the other modelled strategies.

    Table 5

    Clinical effectiveness and cost–utility analysis

    We used data from systematic reviews of glaucoma risk prediction tools and statistical models determining the optimum frequency of monitoring IOP to detect true change from measurement noise to populate the model. For the statistical model we used serial measures of IOP and explored visual field data from the placebo group of randomised controlled treatment trials,11 ,13 which may not be representative of variability in the general population. Repeat measures of visual fields were sparse in these trial data; thus, the model estimates were based solely on IOP data. It should be noted, however, that in the trial cohorts the visual fields (mean deviation) showed no true change over 4 years.8

    We characterised the risk of developing glaucoma using a glaucoma 5-year risk estimator21 and applied this risk prediction model to the simulated cohort. This glaucoma risk model has good predictive utility for estimating the 5-year risk of open-angle glaucoma in an external validation study.19 It is available as an online tool and may be easier to use in clinical practice than risk stratification based on age, CCT and IOP as in the NICE guideline although this has not been evaluated as far as we are aware. In two of our ‘new’ pathways, ocular hypotensive treatment was initiated if the predicted 5-year risk of glaucoma was ≥6%. This cut-off was based on the Ocular Hypertension Treatment Study cohort where low glaucoma 5-year risk was defined arbitrarily as <6%, moderate as 6%–13% and high risk as >13%.22

    We took a 20-year time horizon, which might not be sufficiently long to fully capture the impact on vision in a condition (ie, glaucoma) that is slowly progressive for most people. We chose this horizon as extrapolating 5-year effectiveness data to even longer time horizons would be questionable. If we had taken a lifetime horizon, more cases of visual impairment would have occurred, but given the time it takes to develop visual impairment, discounting would have reduced their impact on any differences between alternative pathways.

    A similar evaluation was conducted in the Netherlands, investigating direct treatment initiation in OHT evaluated from a societal perspective with a lifetime horizon.20 Direct pressure-lowering treatment was less costly and more effective than a strategy of delaying treatment until early signs of glaucoma are apparent. The uncertainty surrounding the model parameters did not affect the conclusions. In the Dutch model, adherence to treatment was not taken into account as it was assumed that it was an implicit part of the treatment effectiveness estimates. However, these estimates were from selected trial populations where adherence may be greater than in clinical practice. We explored varying adherence based on published estimates,23–25 and further data would reduce the uncertainty in our findings.

    Findings were also sensitive to the costs of repeat visits to assess treatment responsiveness (IOP) but were not sensitive to medication costs. For community monitoring we used a cost of £20.70 based upon the NHS sight test tariff. Should this cost be too low, for example should additional charges be incurred for the use of additional tests, then a higher cost would make community monitoring even less cost-effective than the base case results indicate.

    In summary, we find no clear benefit in terms of cost-effectiveness from intensive monitoring of people with OHT to detect glaucoma. Innovative restructuring of the ophthalmic service, which is less costly and has greater emphasis on treatment responsiveness rather than more glaucoma testing, is recommended. The feasibility of alternative and more affordable monitoring pathways should be explored. A rigorously designed prospective comparative study comparing low-intensity surveillance (incorporating alternative treatment thresholds) with current practice is recommended.

    References

    View Abstract

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Collaborators Members of the Surveillance for Ocular Hypertension Study Group: Adriana Botello-Pinzon, Yemisi Takwoingi, Maria Vazquez-Montes, Andrew Elders, Ryo Asaoka, Josine van der Schoot, Cynthia Fraser, Anthony King, Hans Lemij, Roshini Sanders, Stephen Vernon, Aachal Kotecha, Paul Glasziou, David Garway-Heath, David Crabb, Rafael Perera, and Jonathan Deeks.

    • Contributors JMB, AA-B, LV, MR and JAC contributed to the conception and design of the study. JMB, RH, LV, JAC, AA-B, KB and AT contributed to data acquisition. RH developed and conducted the economic model under supervision by LV. JMB drafted the article. All authors contributed to the interpretation of the results, revised the article critically for important intellectual content and gave final approval for publication.

    • Funding This work was part of the Surveillance for Ocular Hypertension study funded by the National Institute for Health Research (NIHR) Health Technology Assessment Programme (07/46/02).

    • Disclaimer The views expressed are those of the authors and not necessarily those of the funding organisation.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement All authors agree to allow review of the data by British Journal of Ophthalmology upon request from the corresponding author.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Linked Articles

    • At a glance
      Keith Barton James Chodosh Jost Jonas