Health Economics Metrics

Topic 54 of 76 · Software engineering and digital delivery

SPACE and DevEx

SPACE (Satisfaction & well-being, Performance, Activity, Communication & collaboration, Efficiency & flow) and DevEx (feedback loops, cognitive load, flow state) are frameworks for measuring developer productivity multi-dimensionally — the field's answer to the discovery that no single metric survives contact with reality.

Why it matters

Both frameworks encode the same hard-won lesson health outcomes research learned decades earlier: a single number (lines of code; blood pressure) misrepresents a multi-dimensional reality, and optimizing it produces gaming, not improvement. SPACE prescribes combining metrics from at least three dimensions, mixing telemetry with self-report — structurally identical to how EQ-5D profiles five dimensions before any index is computed, and why PROMs exist alongside clinical measures. Satisfaction/well-being isn't soft garnish either: it feeds workforce retention economics, where attrition is priced in months of loaded salary.

The math

Neither framework is a formula; both are measurement designs:

SPACE rule: ≥ 3 dimensions, ≥ 1 perceptual (survey) + ≥ 1 system (telemetry) metric

DevEx dimensions and example pairings:
  feedback loops  → CI duration (telemetry) + "waiting feels slow" (survey)
  cognitive load  → docs findability, onboarding time + perceived effort
  flow state      → meeting/interrupt density + self-reported focus

Derived indices (e.g., DX's DXI) map survey composites to time:
vendor claim ≈ 13 min/dev/week per index point — treat as a vendor
benchmark to validate locally, not a constant of nature.

Worked example

A platform team justifies a DevEx investment (CI speedup + docs overhaul) for 300 developers:

Baseline: CI p75 = 28 min; survey "I lose focus waiting for builds": 62% agree
After:    CI p75 = 9 min;  agreement 24%

Time reclaimed (telemetry): 6 builds/day × 19 min × 0.4 usable = ~45 min/day/dev
Capacity value: 300 × 0.75h × 220d × £60/h ≈ £2.97M/year (non-cash-releasing —
see cash-releasing-vs-non-cash-releasing.md; the 0.4 usability factor is the
fragmentation discount from practitioner-time.md)
Perceptual corroboration is what makes the telemetry claim credible — either
alone is gameable; together they triangulate.

Software engineering connection

This doc is the software side; the transfer runs toward health economics. A "quality-adjusted engineer year" — time weighted by a standardized experience index — is the QALY's construction applied to engineering capacity, and it inherits the QALY's rules: weights from a validated instrument (consistent survey, published scoring), elicited before the comparison, never tuned to flatter a favored tool. The SF-6D vs EQ-5D lesson also applies: different instruments give systematically different numbers, so never compare DevEx indices across vendors' instruments.

Pitfalls

  • Single-metric collapse: dashboards that reduce SPACE to one score recreate the problem the framework exists to prevent.
  • Activity metrics as outcomes: commits, PRs, and story points are Activity — the dimension SPACE explicitly warns is most gameable (health analogue: counting procedures, not recoveries).
  • Survey fatigue and Hawthorne effects: quarterly light-touch instruments beat weekly interrogation.
  • Comparing teams: like hospital league tables without case-mix adjustment — context differences (domain, legacy load, on-call) dominate.

Sources