Health Economics Metrics

Topic 13 of 76 · Outcome measures

EQ-5D

EQ-5D is the EuroQol group's standardized questionnaire for measuring health-related quality of life. It is the instrument that produces the utility weights inside most QALY calculations — NICE's reference case names it the preferred measure for adults.

Why it matters

Any digital health product that wants to claim QALYs needs utilities from a validated instrument, and EQ-5D is the default in the UK and much of Europe. It is short enough to embed in an app (5 questions + a visual scale), which means software products can collect HTA-grade outcome data as a side effect of normal use — a structural advantage over drugs, which need dedicated studies.

The math

The EQ-5D-5L asks one question in each of 5 dimensions — mobility, self-care, usual activities, pain/discomfort, anxiety/depression — each answered at 5 levels (no problems … extreme problems), plus a 0–100 visual analogue scale (EQ VAS).

Health state = 5-digit profile, e.g. "21221"
Utility index = value_set(profile)

The value set is country-specific, derived from time-trade-off /
discrete-choice surveys of the general public. Anchors: 1 = full health,
0 = dead; states worse than death are negative (UK 3L set floor: −0.594).

QALY arithmetic then proceeds as duration × utility.

Worked example

A musculoskeletal rehab app measures EQ-5D-5L at onboarding and at 6 months for 1,000 completing users.

Mean utility at baseline:  0.62
Mean utility at 6 months:  0.71
Gain sustained (assume) 1 year: (0.71 − 0.62) × 1.0 = 0.09 QALYs per user

Against a control-group change of 0.03 (natural recovery), the attributable gain is 0.06 QALYs/user. Monetized at £20,000–£30,000/QALY: £1,200–£1,800 of health value per completing user — the number that anchors the app's price negotiation with a payer. (Minimal clinically important differences for the EQ-5D index are commonly in the 0.03–0.08 range, so 0.06 is plausible but must clear the control comparison; see patient-reported outcomes.)

Software engineering connection

  • Instrument it. EQ-5D at signup and at follow-up intervals is a few screens of UI; the payoff is HTA-grade evidence. Get licensing from EuroQol (required, free for some uses).
  • Use the right value set for the deployment country — the same answers score differently in the UK vs Germany vs Japan.
  • Design lesson: EQ-5D shows how a tiny standardized survey plus a published scoring function yields a comparable single index. That's the pattern for any credible developer-experience index too — standardized instrument, published weights, not ad-hoc vibes. See SPACE and DevEx.

Pitfalls

  • Before/after without a comparator — regression to the mean and natural recovery inflate naive gains.
  • Survivorship bias: measuring only users who stayed engaged (see retention and churn).
  • Mixing 3L and 5L versions or value sets across studies — systematically different numbers.
  • Ceiling effects in mildly affected populations: many users score near 1.0 at baseline, leaving no headroom to demonstrate gain.

Sources