Proportional Shortfall

Proportional Shortfall in Severity Assessment

Proportional shortfall measures how severe a condition is by asking what fraction of the health people could still expect is taken away by it. A value of 0.5 means that people with the condition, under current care, can expect half the remaining quality-adjusted life of otherwise similar people without it, and a value of 1 means they lose all of it. The measure began in the Netherlands, where the National Health Care Institute (Zorginstituut Nederland) uses it to set the maximum reference value per QALY gained, and NICE now reads it alongside absolute shortfall when deciding whether QALY gains attract a severity weight. Because each loss is scaled by what the group had left, the ratio treats age very differently from a count of QALYs lost, and that difference drives most of the debate about it.

A ratio of health lost to health expected

The NICE manual, PMG36 (section 6.2.15), describes proportional QALY shortfall as the proportion of future health, including quality and length of life, lost by people living with the condition. It is calculated by dividing the absolute QALY shortfall by the remaining QALYs of a general population with the same age and sex distribution.

$$\text{PS} = \frac{Q_h-Q_c}{Q_h}$$

where PS is the proportional shortfall, $Q_h$ is the expected remaining QALYs of a general population matched on age and sex, and $Q_c$ is the expected remaining QALYs of people with the condition under current care. The numerator, $Q_h-Q_c$, is the absolute shortfall.

The NICE Decision Support Unit guide to severity, Technical Support Document 23 (TSD 23), states that the ratio should normally lie between 0 and 1. It equals zero when the condition reduces neither quality nor length of life and 1 when the condition causes immediate death, and could in theory exceed 1 if quality of life fell below zero for long enough. How $Q_h$ and $Q_c$ are built from life tables, population utilities and the comparator model is set out on the QALY shortfall page.

Origin in the Dutch necessity criterion

Proportional shortfall was proposed as an equity concept before it became a policy tool. Stolk and colleagues, writing in 2004, described it as the ratio of QALYs lost over the QALYs remaining and argued that it combines elements of two popular but conflicting notions of equity, fair innings and severity of illness. Their tentative equity adjustment lowered the cost-effectiveness threshold when a condition was relatively mild.

Van de Wetering and colleagues place the concept within the Dutch criteria for the basic benefits package proposed in 1991, of which necessity had resisted measurement, and proportional shortfall was introduced to give it a quantitative definition. Fair innings would favour people who can expect less lifetime health in total, and prospective health those with the worst outlook if untreated, so the authors present proportional shortfall as an intermediate position. They note that it is not derived from a particular theory of distributive justice and that the Dutch choice was partly pragmatic.

How the ratio treats age and sex

The ratio scales each loss by what the group had left, so very different absolute losses can share one value. Van de Wetering and colleagues give the example of a young patient with a normal expectation of 40 QALYs who loses 20 and an older patient with an expectation of 2 who loses 1. Both have a proportional shortfall of 0.5 and receive the same equity weight.

Every condition causing imminent death therefore scores 1, whatever the age, which the authors call perhaps the most counterintuitive implication of the principle. The measure is indifferent between a 3-year-old losing 80 years and an 80-year-old losing 3, although many people judge treatment more necessary in the first case. Women's higher life expectancy also means that an absolute QALY loss at a given age carries more weight for men than for women.

NICE appraisal data show the pattern in practice. Wailoo and colleagues found that proportional shortfall was broadly flat at around 0.55 where patient starting ages were below 40 years and peaked soon after 60. The age relationship was much weaker than for absolute shortfall, and up to age 55 any group meeting the 1.2 level on proportional shortfall would also meet it on absolute shortfall.

Two ways of turning the ratio into a decision

Health technology assessment bodies use the ratio in two equivalent ways. Van de Wetering and colleagues show that letting the acceptable cost per QALY vary with severity is the same as keeping one threshold and attaching equity weights to the QALY gains.

$$\frac{\Delta C}{w \, \Delta Q} \le \lambda \quad \Longleftrightarrow \quad \frac{\Delta C}{\Delta Q} \le w \lambda$$

where $\Delta C$ is the incremental cost of the new technology, $\Delta Q$ is its incremental QALY gain over current care, $w$ is the severity weight and $\lambda$ is the cost per QALY accepted for a condition with no extra weight.

The Netherlands varies the threshold. The National Health Care Institute's 2024 cost-effectiveness assessment framework links three classes of proportional shortfall to maximum reference values for the additional cost per QALY, and describes each as an upper value, not a target. Reckers-Droog and colleagues report that technologies below 0.10 are in principle not reimbursed, and that the institute formalised this use of proportional shortfall, with absolute shortfall considered at the appraisal stage, in 2018.

NICE instead weights the QALYs. Table 6.1 of PMG36 gives a weight of 1.2 for a proportional shortfall from 0.85 to 0.95 and 1.7 for at least 0.95, alongside absolute shortfall cut-offs of 12 and 18 QALYs. The committee applies whichever measure implies the greater severity, and a value exactly on a cut-off takes the higher level.

Worked example with one ratio under two rules

All numbers in this example are illustrative and not taken from any appraisal or dossier. The same discounted values are used under both rules for simplicity, although jurisdictions differ in whether they discount the shortfall. Population A is elderly and population B is young.

1. Population A. The matched general population expects 9.0 QALYs and patients under current care expect 0.9.

$$\text{PS}_A = \frac{9.0-0.9}{9.0} = \frac{8.1}{9.0} = 0.90$$

where the absolute shortfall is 8.1 QALYs and the proportional shortfall is 0.90.

2. Population B. The matched general population expects 24.0 QALYs and patients under current care expect 12.0.

$$\text{PS}_B = \frac{24.0-12.0}{24.0} = \frac{12.0}{24.0} = 0.50$$

where the absolute shortfall is 12.0 QALYs and the proportional shortfall is 0.50. Population B loses more QALYs than population A but a smaller share of its remaining health.

3. The Dutch rule. Population A falls in the 0.71 to 1.00 class, with a reference value of EUR 80,000 per QALY, and population B in the 0.41 to 0.70 class, with EUR 50,000. An ICER of EUR 60,000 per QALY would sit below the reference value for A and above it for B.

4. The NICE rule. For population A, 0.90 lies in the 0.85 to 0.95 band and an absolute shortfall of 8.1 lies below 12, so the weight is 1.2. For population B, 0.50 earns no weight, but the absolute shortfall of 12.0 sits exactly on the cut-off, so the weight is again 1.2. If a technology cost GBP 30,000 more than current care and gained 0.75 QALYs, the unweighted ICER would be 30,000 / 0.75 = GBP 40,000 per QALY. The weighted gain would be 0.75 × 1.2 = 0.9 QALYs, giving 30,000 / 0.9 = GBP 33,333 per QALY, inside the GBP 25,000 to GBP 35,000 per QALY range that PMG36 (sections 6.3.4 to 6.3.7) uses for committee judgements.

5. A worse outlook for population A. Suppose current care gave population A only 0.4 QALYs.

$$\text{PS}_A = \frac{9.0-0.4}{9.0} = \frac{8.6}{9.0} = 0.956$$

where 0.956 is at least 0.95, so the NICE weight would rise to 1.7, while the Dutch class would not change.

The Dutch rule separates the two populations sharply, while the NICE rule reaches the same weight for both through different measures.

Evidence from appraisal practice

Dutch and English data show where the ratio and the QALY count part company. Reckers-Droog and colleagues analysed 56 of the 57 dossiers appraised by the Dutch advisory committee between May 2018 and April 2026. A cluster of 28 had a high proportional shortfall (mean 0.88) but a relatively low absolute shortfall (mean 14.28 QALYs), and 27 of the 28 concerned oncology drugs for older patients. In one dossier, a proportional shortfall of 0.99 with an absolute shortfall of 70 QALYs led the committee to judge an ICER above the EUR 80,000 reference value acceptable, subject to price negotiations.

In the NICE samples analysed by Wailoo and colleagues, mean proportional shortfall in the primary sample was 0.62, and 6.9 per cent of decisions reached the 1.7 level on it. The end-of-life criteria that the severity modifier replaced were more strongly associated with high proportional shortfall (correlation 0.51) than with high absolute shortfall (0.31).

Choices that change the ratio

A reported proportional shortfall reflects normative and data choices as well as clinical evidence. PMG36 asks that the data focus on the specific population for which the technology will be used and that both shortfalls be discounted at the reference-case rate, which TSD 23 gives as 3.5 per cent. The Dutch institute's 2018 report on calculating disease burden asks for subgroup estimates where relevant, since an average can hide a small group with a very severe loss.

Van de Wetering and colleagues show how prevention depends on such choices. Averaging over everyone treated gives a very low ratio, because few would have developed the disease, and the Dutch position is to calculate it in the subgroup that would. For a risk factor that, untreated, causes death 20 years later and shortens life by 10 years, counting from the start of treatment gives about 0.33 and counting from the onset of illness gives 1. Episodic conditions raise a similar problem, since a stable mild loss and a short severe episode averaged over a year can give the same small number.

Proportional, absolute and QALY shortfall

Proportional shortfall is one of two readings of the same gap between expected health with and without a condition. The proportional reading asks how much of the future is lost rather than how many QALYs, and so tends to rank conditions in older people higher.

It is distinct from age weighting, which values QALYs by the age of the person receiving them, and from distributional cost-effectiveness, which tracks how health gains and losses fall across social groups. A severity weight based on the ratio is one form of equity-weighted analysis, and it carries an opportunity cost in health for patients whose care is displaced.

Criticisms and misreadings

Support for proportional shortfall as a description of public values is mixed. Reckers-Droog and colleagues concluded in 2018 that empirical support was limited and that the measure may insufficiently reflect societal preferences regarding age and the reduction of lifetime health inequalities. In their 2019 survey of 1,025 members of the Dutch public, respondents gave a higher weight to younger patients irrespective of severity when severity and age were varied together. The Dutch institute's 2018 report still judged proportional shortfall the most just way to measure disease burden, while noting that it does not distinguish between age groups, and began reporting its separate elements, including absolute shortfall, to keep that consequence visible.

Two misreadings recur.

1. Treating a high ratio as a large loss. Among the Dutch dossiers with a proportional shortfall between 0.5 and 1.0, absolute shortfall varied between around 10 and more than 50 QALYs.
2. Reading it as lifetime disadvantage. TSD 23 notes that the measure looks forward from the point the technology is considered, so health already lost does not count.

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. Sections 6.2.12 to 6.2.21, Table 6.1 and sections 6.3.4 to 6.3.7. https://www.nice.org.uk/process/pmg36/chapter/committee-recommendations-2
• Wailoo A. NICE DSU Technical Support Document 23: A guide to calculating severity shortfall for NICE evaluations. NICE Decision Support Unit, University of Sheffield. January 2024, updated September 2026. https://sheffield.ac.uk/nice-dsu/tsds/severity-shortfall-tsd
• Stolk EA, van Donselaar G, Brouwer WB, Busschbach JJ. Reconciliation of economic concerns and health policy: illustration of an equity adjustment procedure using proportional shortfall. PharmacoEconomics. 2004;22(17):1097-1107. doi:10.2165/00019053-200422170-00001
• van de Wetering EJ, Stolk EA, van Exel NJ, Brouwer WB. Balancing equity and efficiency in the Dutch basic benefits package using the principle of proportional shortfall. European Journal of Health Economics. 2013;14(1):107-115. doi:10.1007/s10198-011-0346-7
• Zorginstituut Nederland. Ziektelast in de praktijk: de theorie en praktijk van het berekenen van ziektelast bij pakketbeoordelingen. 7 May 2018. https://www.zorginstituutnederland.nl/documenten/2018/05/07/ziektelast-in-de-praktijk
• Zorginstituut Nederland. Beoordelingskader kosteneffectiviteit van zorg. 26 November 2024. https://www.zorginstituutnederland.nl/documenten/2024/11/26/beoordelingskader-kosteneffectiviteit-van-zorg
• Reckers-Droog V, Redeker S, Busschbach J, Vijgen S, Link A, Timmers L. Using proportional and absolute shortfall in reimbursement decision-making in the Netherlands: implications for oncology drugs for older patients. PharmacoEconomics. 2026;44(8):899-906. doi:10.1007/s40273-026-01626-z
• Reckers-Droog VT, van Exel NJA, Brouwer WBF. Looking back and moving forward: on the application of proportional shortfall in healthcare priority setting in the Netherlands. Health Policy. 2018;122(6):621-629. doi:10.1016/j.healthpol.2018.04.001
• Reckers-Droog V, van Exel J, Brouwer W. Equity weights for priority setting in healthcare: severity, age, or both? Value in Health. 2019;22(12):1441-1449. doi:10.1016/j.jval.2019.07.012
• Wailoo A, Bell Gorrod H, Dunning L, Kenny J, Leckenby E, Shah K. The NICE experience of designing and utilising severity weights. Health Policy OPEN. 2026;10:100156. doi:10.1016/j.hpopen.2025.100156