Reference Case Requirement

Reference Case Requirement in NICE and Other HTA Methods

A reference case is built from separate requirements, each of which fixes one methodological choice that would otherwise be left to the analyst, such as whose costs count or what discount rate applies. Treating each requirement as a unit matters in practice because a submission can follow most of a reference case and depart from one part of it, and an appraisal committee then judges that departure on its own terms. NICE sets out its requirements row by row in table 4.1 of its technology appraisal manual (PMG36), where the discounting row asks for the same annual rate for costs and health effects, currently 3.5%. Requirements also change one at a time, as the adult EQ-5D row did when NICE issued its interim methods statement PMG51 in August 2026.

What one requirement fixes and why it stands alone

A reference case exists so that appraisals of different technologies can be compared, which needs the same methods for the parts of an analysis that do not depend on the technology. Each requirement names one element of the analysis and states the method the body expects for that element. In PMG36 table 4.1 each row pairs an element with the reference-case method and points to the section of the manual that explains it in detail.

Seen as separate units, requirements differ in what they act on. Some decide which effects enter the analysis at all, as the perspective on costs and on outcomes does. Others decide how an included effect is turned into a number, such as the instrument used to measure health-related quality of life or the discount rate applied to future costs and QALYs. A third group fixes the frame of the analysis, as the time horizon does. Because each requirement touches a different part of a cost-utility model, a change to one can move the ICER while leaving every other input untouched.

Examples from table 4.1 of PMG36

PMG36, the NICE technology appraisal and highly specialised technologies guidance manual, gives its reference case in section 4.2 as a table. Four of its rows show the range of elements a single requirement can cover.

The text behind the table adds detail to several rows. Section 4.3.4 says the valuation of health-related quality of life measured by patients or their carers should be based on public preferences from a representative sample of the UK population, using a choice-based method. A requirement is therefore more than its table cell, and the section the row points to often holds the conditions that decide whether a method complies.

How strongly a requirement binds

Requirements are not all worded with the same force, and reading that force correctly decides whether an alternative method counts as a departure. Three strengths can be told apart in NICE's text, although NICE does not label them this way.

Some requirements state the method that should be used. The cost perspective and the 3.5% discount rate are of this kind, and an analysis that uses a different cost perspective or a different rate is outside the reference case. Other requirements are expressed as a preference, as when the EQ-5D is named as the preferred measure in adults. A preference leaves room for another method, but only on evidence. PMG36 says that in some circumstances the EQ-5D may not be the most appropriate measure, and that a case for this needs qualitative empirical evidence that important dimensions of health are missing. When EQ-5D data are not available, PMG36 allows utilities to be estimated by mapping from other measures, and it treats this as a departure from the reference case.

The third strength covers methods that a body accepts only alongside a "should" rule, never in place of it. Discounting shows the pattern. The 3.5% rate stays the reference case, but PMG36 (section 4.5.3) allows the committee to consider a non-reference-case rate of 1.5% a year for both costs and health effects when three conditions hold. The technology is for people who would otherwise die or have a very severely impaired life, it is likely to restore them to full or near-full health, and the benefits are likely to be sustained over a very long period. The committee must also be satisfied that there is a highly plausible case for the maintenance of benefits, and the 1.5% analysis is then weighed beside the 3.5% one.

Departures, justification and sensitivity analysis

A departure from one requirement does not remove an analysis from consideration. PMG36 allows additional analyses whose methods differ from the reference case, provided the differences are justified and the results are clearly distinguished from the reference-case results. Keeping the two sets of results apart lets the committee see what the change of method did.

In practice a departure is handled in two steps. The submission states the reason for departing, such as evidence that the preferred instrument misses an important dimension of health in the condition. It then shows how much the departure changes the result, usually as a sensitivity analysis or scenario analysis that reruns the model with the reference-case method and with the alternative. Where the departure is forced by missing data rather than chosen, as with mapping to EQ-5D utilities, the uncertainty that the mapping adds becomes part of the decision.

When a requirement changes, the EQ-5D in 2026

The adult EQ-5D requirement shows how one row of a reference case can change while the others stay fixed. NICE published its interim methods statement, PMG51, on 27 August 2026. It makes the EQ-5D-5L descriptive system and the new UK value set NICE's preferred approach to measuring and valuing health-related quality of life in adults, and says the UK 5L value set, published by Rowen and colleagues in 2026, should be used to derive all adult utility values in reference-case analyses. NICE does not recommend a specific measure of health-related quality of life for children and young people.

The change does not reach back to work already under way, since NICE has said the EQ-5D-5L value set will be used for all new topics starting after the interim statement was published. PMG36 was last updated on 13 October 2026, for a change to the section on varying the funding requirement to take account of net budget impact, and its EQ-5D sections, including table 4.1, are among those due for update. Until that revision is published, the manual and the interim statement have to be read together. The discount rate, perspective and time horizon rows were left as they were, which is why the EQ-5D-5L value set change moved utility values without touching any other part of the reference case.

The discount rate under three reference cases

Most HTA bodies with an economic reference case cover the same elements, but the rule attached to each element differs. The discount rate shows this most plainly, because the element is identical everywhere and only the number changes.

ICER, the Institute for Clinical and Economic Review in the United States, sets its rules in ICER's Reference Case for Economic Evaluations, which discounts costs and outcomes at 3% a year from a default health care system perspective. Canada's Drug Agency (CDA-AMC, formerly CADTH) sets its reference case in the fourth edition of the Guidelines for the Economic Evaluation of Health Technologies, published in 2017 under the CADTH name. Costs and outcomes beyond one year are discounted at 1.5% a year, a figure the guidelines link to the long-term cost of borrowing for Canadian provinces, and non-reference-case analyses use 0% and 3%.

Illustrative worked example of one requirement, the discount rate

All numbers in this example are illustrative. A one-off preventive treatment for people who are currently well costs GBP 36,000 per patient, paid now, and yields a gain of 2 QALYs that arrives as a single block 10 years later. Costs incurred now are not discounted, so only the QALY gain changes with the discount rate.

$$Q_{PV} = \frac{Q_t}{(1+r)^t}$$

where $Q_{PV}$ is the present value of the QALY gain, $Q_t$ is the undiscounted gain received in year $t$, $r$ is the annual discount rate and $t$ is the number of years until the gain arrives.

1. NICE reference case at 3.5%. The divisor is 1.035 raised to the power 10, which is 1.4106. The discounted gain is 2 ÷ 1.4106 = 1.4178 QALYs, and 36,000 ÷ 1.4178 = 25,391, a cost of GBP 25,391 per QALY.

2. NICE non-reference case at 1.5%. The divisor is 1.015 raised to the power 10, which is 1.1605. The discounted gain is 2 ÷ 1.1605 = 1.7234 QALYs, and 36,000 ÷ 1.7234 = 20,889, a cost of GBP 20,889 per QALY.

3. The same model at the 3% rate used by the Institute for Clinical and Economic Review. The divisor is 1.03 raised to the power 10, which is 1.3439. The discounted gain is 2 ÷ 1.3439 = 1.4882 QALYs, and 36,000 ÷ 1.4882 = 24,190, a cost of GBP 24,190 per QALY.

4. No discounting. The gain stays at 2 QALYs, and 36,000 ÷ 2 = 18,000, a cost of GBP 18,000 per QALY.

Only one requirement changed between these runs, yet the cost per QALY lies between GBP 18,000 and GBP 25,391. Against an illustrative threshold of GBP 25,000 per QALY, the lower end of the range NICE has applied since April 2026, the treatment narrowly fails under the NICE reference case and passes under the 1.5% analysis. In a NICE appraisal the 3.5% figure would be the reference-case result, and the 1.5% figure could be weighed only if the treatment met the three conditions in section 4.5.3, which a preventive treatment for people who are currently well would not meet. The size of the gap comes from the 10-year delay, so requirements on discounting matter most for technologies whose benefits arrive long after their costs.

Requirement, reference case and base case

A reference case requirement is easily confused with nearby terms. The reference case is the whole set of requirements, and a single requirement is one member of that set. A base case is the analyst's preferred set of inputs and assumptions for a particular model. A base case can follow every reference-case requirement or depart from some, and the requirements decide which of its methods count as reference case.

A requirement also differs from a decision rule. NICE's reference case says how costs and QALYs are to be estimated, while the threshold and the factors a committee weighs in reaching a recommendation sit in other parts of the manual. A requirement on perspective decides which costs are counted, but it says nothing about whether the resulting ICER is acceptable.

Common misreadings of reference case requirements

The most frequent error is to treat a requirement as a ban on every other method. PMG36 lets a committee consider analyses that depart from the reference case, provided the departure is justified and the results are shown separately. A justified departure still is not a reference-case analysis, however, and the reference-case result keeps its place.

Reading one requirement without its date is a second error, because a single row can be replaced while the manual's text waits for revision, as the adult EQ-5D row shows. A third error is carrying a requirement from one jurisdiction into another. A model built for CDA-AMC at 1.5% and resubmitted to NICE without changing the rate reports a lower cost per QALY than the NICE reference case would give for the same evidence whenever the benefits arrive later than the costs, as the worked example shows.

Sources

• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 31 January 2022, last updated 13 October 2026. Section 4.2 (table 4.1), sections 4.3 and 4.5. https://www.nice.org.uk/process/pmg36/chapter/economic-evaluation-2
• National Institute for Health and Care Excellence. Interim methods statement: implementing the EQ-5D-5L value set (PMG51). Published 27 August 2026. https://www.nice.org.uk/process/pmg51/chapter/statement-on-using-the-eq-5d-5l-value-set
• National Institute for Health and Care Excellence. Changes to NICE's cost-effectiveness thresholds take effect. News article, April 2026. https://www.nice.org.uk/news/articles/changes-to-nice-s-cost-effectiveness-thresholds-take-effect
• Institute for Clinical and Economic Review. ICER's Reference Case for Economic Evaluations. Boston: ICER; current as of 23 October 2025. https://icer.org/wp-content/uploads/2024/02/ICER_Reference-Case_For-Publication_070126.pdf
• CADTH. Guidelines for the Economic Evaluation of Health Technologies: Canada. 4th edition. Ottawa: CADTH; 2017. https://www.cda-amc.ca/guidelines-economic-evaluation-health-technologies-canada-4th-edition