Work Productivity and Activity Impairment questionnaire (WPAI)
In plain English
Work Productivity and Activity Impairment questionnaire (WPAI) is a multi-item scale (6 items). What it claims to measure: Health-related impairment of work and of regular daily activities over the past seven days, decomposed into absenteeism (work time missed), presenteeism (reduced effectiveness while working), overall work productivity loss, and non-work activity impairment. Licence for an employer or vendor: could not be verified against the steward's current terms; no class is asserted.
The published evidence is moderate for convergent and discriminant validity and responsiveness to change; weak for test-retest reliability. No published evidence was located for criterion validity against organisational outcomes (absence, turnover, performance) and measurement invariance: the registry searched and found none in the sweep to date, so if you need those properties evidenced, this instrument does not yet carry them. Criterion validity against a reference standard does not apply to this construct: no reference standard exists for it, so nothing is graded and nothing is missing. All 3 graded properties rest on evidence from clinical, student or otherwise non-working samples; where the property is sensitive to population that evidence cannot carry a High grade, and the reason is stated on each cell below.
Grades summarise the published evidence for this instrument on its own terms; they are not comparable across instruments, and this page makes no recommendation. Whether an instrument fits your workforce is a judgement this registry informs but cannot make. Full evidence, with citations, below. This summary is generated from the record's data, not written by hand.
Identity
Version: General health version WPAI:GH V2.0 is the generic form; numerous disease-specific adaptations exist (WPAI:CD, WPAI:PsO, WPAI-Lupus v2.0, WPAI:GERD, WPAI:AS and others). The original instrument was published in 1993.
Structure: 6 items (recall period one week): whether currently employed, hours missed due to the health problem, hours missed for other reasons, hours actually worked, a 0 to 10 rating of the degree the health problem affected productivity while working (presenteeism), and a 0 to 10 rating of the degree it affected regular non-work activities. Four scores are derived as percentages: absenteeism, presenteeism, overall work productivity loss (absenteeism plus presenteeism), and activity impairment.
Original citation: Reilly MC, Zbrozek AS, Dukes EM. The validity and reproducibility of a work productivity and activity impairment instrument. PharmacoEconomics 1993;4(5):353-365. https://doi.org/10.2165/00019053-199304050-00006
Steward / publisher: Reilly Associates Health Outcomes Research (Margaret Reilly), distributing via reillyassociates.net; translations and licensed disease-specific derivatives are coordinated through Mapi Research Trust / ePROVIDE.
Licence status: not yet verified against current steward terms. The steward is Reilly Associates (reillyassociates.net), with translations and disease-specific derivatives coordinated via Mapi Research Trust / ePROVIDE, but the steward's current licence terms could not be read at the last verification attempt: direct fetch of the licence pages (wpai_general.html and wpai_gh.html) returned HTTP 403, and web search returned only page titles and URLs, not the licence body text. Per schema rule 7, no licence terms are asserted here from literature or background knowledge. The current distribution terms must be read directly from reillyassociates.net (or obtained from Mapi Research Trust) before any licence claim is relied upon. It should NOT be described as an open, CC or public-domain licence absent such verification.
Constructs claimed
Health-related impairment of work and of regular daily activities over the past seven days, decomposed into absenteeism (work time missed), presenteeism (reduced effectiveness while working), overall work productivity loss, and non-work activity impairment. It presupposes the presence of a health problem and quantifies its productivity impact rather than measuring wellbeing or job satisfaction directly.
Evidence
Structural validity Not applicable (category difference)untestedevidence form: canonical
absence type: category-error the property does not apply to this construct
The WPAI is not a latent multi-item scale; it yields four derived percentage scores from six administrative and single-rating items, so a conventional common-factor structure does not apply and formal factor-analytic evidence is essentially absent. The original development framed the scores as directly computed indices of work and activity impairment rather than reflective indicators of a latent trait (Reilly 1993), and reviews of self-report productivity instruments describe the WPAI in the same terms (Prasad 2004). Because presenteeism and activity impairment are each single-item 0 to 10 ratings and absenteeism is arithmetically derived from reported hours, structural validity in the factor-analytic sense is a partial category mismatch for this instrument and no credible confirmatory factor evidence was located.
Convergent and discriminant validity Moderatewell-establishedevidence form: canonical
indirect rheumatoid arthritis patients, axial spondyloarthritis, Crohn's disease and lupus samples; no working-adult or general-population sample (flag basis: rheumatoid arthritis patients; axial spondyloarthritis patients; Crohn's disease patients; lupus patients)
Convergent validity is the WPAI's best-evidenced property and is consistently supported. In the original study the WPAI impairment measures correlated positively with established health-perception, role, pain and symptom-severity measures, and the validation set explained 54 to 64 per cent of the variance in productivity and activity scores (Reilly 1993). In rheumatoid arthritis patients, absenteeism correlated moderately (r = 0.34 to 0.39) and presenteeism and activity impairment strongly (0.67 to 0.77) with function, pain, fatigue and disease severity, and the scores distinguished patients by health status (Zhang 2010). In axial spondyloarthritis patients, presenteeism, overall work productivity loss and activity impairment correlated strongly (r greater than 0.6) with BASFI, patient global and pain, and the instrument discriminated between disease-activity groups (Phang 2020). In Crohn's disease patients the domains discriminated across CDAI, SF-36 PCS and MCS, IBDQ and EQ-VAS strata (Reilly 2008), and in lupus patients the scores correlated with SLEDAI-2K and with LupusQoL domains (activity impairment approximately -0.52 to -0.58 with physical health, pain and planning) (Nieto 2026). Correlations are typically moderate to strong for presenteeism, overall loss and activity impairment, and weaker for absenteeism, which is often floor-loaded. Sample sizes were 106 employed individuals with a health problem (Reilly 1993), 150 rheumatoid arthritis patients, in whom absenteeism correlated 0.34 to 0.39 and presenteeism and activity impairment 0.67 to 0.77 with function, pain, fatigue and disease severity (Zhang 2010), 168 axial spondyloarthritis patients (Phang 2020), 662 Crohn's disease trial patients, where the worst CDAI stratum showed 10.5 per cent more work impairment and 10.4 per cent more activity impairment than the best (Reilly 2008), and 1,083 Latin American lupus patients, with presenteeism and overall work impairment correlating about -0.49 to -0.53 with physical health and pain (Nieto 2026).
Criterion validity: reference standard Absent (the property does not apply to this construct)untestedevidence form: canonical
absence type: category-error the property does not apply to this construct
There is no diagnostic or health gold-standard reference against which the WPAI has been criterion-validated; the construct (health-related productivity loss) has no external diagnostic criterion. Validation studies test the WPAI against other self-report health and disease-activity measures, which is convergent rather than criterion evidence (Reilly 1993, Prasad 2004). No true reference-standard criterion validity was located in the latest review pass.
Criterion validity: organisational Absent (searched; none found in the sweep to date)untestedevidence form: canonical
absence type: population-general searched; none found in the sweep to date
Criterion validity against objective organisational work outcomes (employer-recorded sickness absence, payroll or timekeeping data, objective output, turnover) was not established in the sources retrieved in the latest review pass. Validation has relied on self-reported health status, disease-activity indices and health-related quality of life as comparators rather than on independent, objectively measured work outcomes (Reilly 1993, Zhang 2010, Prasad 2004). This is an important and honestly-reported gap: the instrument's productivity scores are self-report and have not, in the evidence located, been benchmarked against objective absence or performance records. The absence of objectively-anchored organisational criterion evidence is itself the finding.
Internal consistency Not applicable (category difference)untestedevidence form: canonical
absence type: category-error the property does not apply to this construct
Internal consistency has limited meaning for the WPAI because its domains are single-item ratings or arithmetically derived indices rather than sets of items tapping one latent trait, so Cronbach's alpha across the domains is of questionable applicability. Where alpha has been reported it has been low: in a Latin American lupus cohort the values were 0.54 (absenteeism), 0.64 (presenteeism), 0.62 (overall work impairment) and 0.62 (activity impairment) (Nieto 2026). These low values are consistent with the instrument's structure rather than indicating a defective scale. Under rubric 1.4 internal consistency is a property of a multi-item scale; the WPAI has none, so the cell is Not-applicable (category-error) and the reported alphas stand in the findings for the reader without a grade.
Test-retest reliability Lowthinevidence form: canonical
indirect axial spondyloarthritis patients and migraine trial cohorts; no working-adult or general-population sample (flag basis: Axial spondyloarthritis patients without treatment change; Episodic or chronic migraine, multinational trial)
| Coefficient | Type | Interval | Sample | Population | Evidence form |
|---|---|---|---|---|---|
| 0.54 | ICC | 2 weeks | 50 | Axial spondyloarthritis patients without treatment change, Singapore (English-speaking) | canonical |
| 0.76 | ICC | 2 weeks | 50 | Axial spondyloarthritis, Singapore (presenteeism) | canonical |
| 0.79 | ICC | 2 weeks | 50 | Axial spondyloarthritis, Singapore (activity impairment) | canonical |
| 0.83 | ICC | 2 weeks | 50 | Axial spondyloarthritis, Singapore (overall work productivity loss) | canonical |
| 0.360 | ICC | about 3 months | stable subgroup within 444 analysed | Episodic or chronic migraine, multinational trial (overall work productivity loss) | canonical |
| 0.438 | ICC | about 3 months | stable subgroup within 444 analysed | Episodic or chronic migraine (presenteeism) | canonical |
| 0.446 | ICC | about 3 months | stable subgroup within 444 analysed | Episodic or chronic migraine (non-work activity impairment) | canonical |
Formal test-retest evidence is thin and comes from disease cohorts. The strongest single dataset (axSpA, n=50, 2 weeks) gives ICCs from 0.54 (absenteeism) to 0.83 (overall work productivity loss), moderate to good. A migraine dataset at a much longer roughly three-month interval gives weaker ICCs of 0.36 to 0.45, but the long interval and expected symptom change make this a lower bound rather than pure unreliability. The original 1993 paper assessed reproducibility across two administration methods (self versus interviewer) rather than classical short-interval test-retest, and reported higher construct validity for interviewer administration. No general-workforce test-retest study was located. Absenteeism is the least reliable domain, consistent with its floor-heavy distribution.
Measurement invariance Absent (searched; none found in the sweep to date)untestedevidence form: canonical
absence type: population-general searched; none found in the sweep to date
No formal measurement-invariance testing (configural, metric or scalar) across sex, age, occupation, language or time was located in the latest review pass. This is expected given that the WPAI is scored as derived indices rather than as a latent multi-item factor, so the usual invariance framework does not directly apply. Cross-cultural adaptation studies exist (for example the Latin American lupus validation, Nieto 2026, and the Singapore axSpA study, Phang 2020), but these establish adaptation and construct validity, not statistical invariance.
Responsiveness and MIC Moderatewell-establishedevidence form: mixed
indirect Crohn's disease, psoriasis and migraine patients; no working-adult or general-population sample (flag basis: Crohn's disease trial patients; psoriasis patients; episodic and chronic migraine patients)
Responsiveness is a genuine strength and is the reason the WPAI is used as a productivity endpoint in clinical trials. In Crohn's disease, WPAI:CD scores discriminated remission from non-remission at 26 weeks, with standardised response means moderate to large in remitters and small in non-remitters (Reilly 2008). In psoriasis patients, minimal clinically important differences for the WPAI-PsO work productivity loss and activity impairment domains were derived from three phase 3 ixekizumab trials (UNCOVER-1/-2/-3, N=3126) using anchor and distribution methods (Wu 2019). In episodic and chronic migraine patients, responders on migraine-day and MSQ anchors showed significantly greater WPAI improvement than non-responders, and a meaningful within-patient change threshold of about 20 percentage points was identified for the domain scores (Ford 2023). Minimal important change values are therefore condition-specific rather than a single generic value. Sample sizes and thresholds behind these estimates: n = 662 Crohn's disease trial patients, with standardised response means below 0.5 in the non-remission group and above 0.5 in remitters (Reilly 2008); the psoriasis MCID was estimated at 20% for both the work productivity loss and activity impairment domains in 3126 patients (792 placebo, 2334 ixekizumab) (Wu 2019); and the migraine threshold of -20 percentage units came from 444 of 462 randomised CONQUER patients who completed the WPAI (Ford 2023).
Populations, languages and norms
The WPAI has been very widely translated and adapted, with disease-specific versions (WPAI:GH general health, WPAI:CD, WPAI:PsO, WPAI-Lupus v2.0, WPAI:GERD, WPAI:AS and others) and numerous language versions distributed through the steward and Mapi Research Trust / ePROVIDE. Validation cohorts span rheumatoid arthritis (Zhang 2010), Crohn's disease (Reilly 2008), axial spondyloarthritis (Phang 2020), psoriasis (Wu 2018), migraine (Ford 2023) and lupus (Nieto 2026). No general-population or working-population normative reference values were located; the instrument reports impairment relative to the individual's own recent week rather than to population norms. A recent UK Occupational Medicine instrument profile discusses its use (Walker-Bone 2025).
Criticisms and controversies
Recurring criticisms: (1) the WPAI is entirely self-report and, in the evidence retrieved, has not been benchmarked against objective absence, payroll or output records, so its organisational criterion validity against real work outcomes is unestablished; (2) it presupposes a health problem and measures health-related productivity loss, making it a poor fit for general wellbeing screening in an unselected workforce; (3) its 'domains' are single-item or derived-percentage scores, so internal-consistency and factor-analytic properties are weak or not meaningfully applicable, and reported alphas are low; (4) the absenteeism domain is floor-heavy and the least reliable and least correlated component; (5) minimal important change values are condition-specific rather than generic, complicating cross-context interpretation; (6) almost the entire psychometric base sits in clinical disease cohorts and pharmaceutical trials, so general-workforce evidence is thin.
References (9)
- Reilly MC; Zbrozek AS; Dukes EM (1993). The Validity and Reproducibility of a Work Productivity and Activity Impairment Instrument https://doi.org/10.2165/00019053-199304050-00006
- Prasad M; Wahlqvist P; Shikiar R; Shih YT (2004). A Review of Self-Report Instruments Measuring Health-Related Work Productivity https://doi.org/10.2165/00019053-200422040-00002
- Reilly MC; Gerlier L; Brabant Y; Brown M (2008). Validity, reliability, and responsiveness of the work productivity and activity impairment questionnaire in Crohn's disease https://doi.org/10.1016/j.clinthera.2008.02.016
- Zhang W; Bansback N; Boonen A; Young A; Singh A; Anis AH (2010). Validity of the work productivity and activity impairment questionnaire - general health version in patients with rheumatoid arthritis https://doi.org/10.1186/ar3141
- Wu J; Lin C; Sun L; Goldblum O; Zbrozek A; Burge R et al (2018). Minimal clinically important difference (<scp>MCID</scp>) for work productivity and activity impairment (<scp>WPAI</scp>) questionnaire in psoriasis patients https://doi.org/10.1111/jdv.15098
- Ford JH; Ye W; Ayer DW; Mi X; Bhandari S; Buse DC et al (2023). Validation and meaningful within-patient change in work productivity and activity impairment questionnaire (WPAI) for episodic or chronic migraine https://doi.org/10.1186/s41687-023-00552-4
- Phang JK; Kwan YH; Fong W; Tan CS; Lui NL; Thumboo J et al (2020). Validity and reliability of Work Productivity and Activity Impairment among patients with axial spondyloarthritis in Singapore https://doi.org/10.1111/1756-185X.13801
- Nieto RE; Hernández L; Quintana R; Fernández‐Ávila D; Santillan EP; Subils G et al (2026). Cross Cultural Validation of Work Productivity and Activity Impairment Questionnaire in Lupus Patients From Latin America https://doi.org/10.1111/1756-185x.70641
- Walker-Bone K (2025). The Work Productivity and Activity Impairment (WPAI) questionnaire https://doi.org/10.1093/occmed/kqaf097
Record notes
Overall confidence: convergent validity and responsiveness are well-established (High) and are the instrument's genuine strengths; test-retest reliability is thin and Low; internal consistency is Low and only partially applicable; structural validity and measurement invariance are largely category mismatches for a derived-index instrument and are Not-applicable/Absent; criterion validity against a diagnostic reference standard is Absent by construct, and criterion validity against objective organisational work outcomes is Absent and thin, which is the most important honest gap for a workplace registry. The evidence base is broad but clinical: nearly all coefficients come from disease cohorts (RA, Crohn's, axSpA, psoriasis, migraine, lupus) in trial settings, so population indirectness for a general working-adult audience is flagged throughout. Nine DOIs, all verified to resolve via Crossref in the latest review pass. LICENCE HONESTY: schema rule 7 could not be fully satisfied because the steward's current licence page (reillyassociates.net) returned HTTP 403 on direct fetch and web search surfaced only page titles; the stated terms reflect the steward's known distribution model, corroborated by the Mapi/ePROVIDE listing, but were not read from the live steward page in the latest review pass, so licence_verified_date is recorded as not verified in the latest review pass. No founding paper or review was used as a licence source.