Maslach Burnout Inventory (MBI, workplace forms)

Structural validity Moderatecontestedevidence form: mixed

indirect most confirmatory evidence is from healthcare, nursing and teaching samples outside the UK, with UK nurses included only within multi-country pooled datasets

A meta-analytic review of 45 exploratory and confirmatory factor studies supported the three-factor structure (Worley J. 2008), and a large eight-country secondary analysis of 54,738 hospital nurses reproduced three factors, though two items (6 and 16, on 'stress' and 'strain' of working with people) consistently cross-loaded onto depersonalisation rather than emotional exhaustion (Poghosyan L. 2009). Factorial invariance work on 2,923 teachers likewise argued for broad three-factor validity while flagging six of the 22 items for content re-examination (Byrne B. 1994). The three-factor model has fitted well in newer confirmatory studies (for example the MBI-GS in 978 Colombian workers, RMSEA 0.05, CFI 0.99; Bravo D. 2021), but a meta-analytic structural equation modelling reanalysis found the data best fitted a bifactor model dominated by a general factor, questioning the independence of the three scales (Aguayo-Estremera R. 2024). The separability of the personal accomplishment/professional efficacy subscale remains the principal structural dispute, and a Vietnamese study reassessed competing 20- and 22-item three-factor models on 1,162 healthcare workers (Bui T. 2022).

Sub-grades (evidence differs by subgroup):

• {"subgroup": "Emotional exhaustion / depersonalisation", "grade": "Moderate", "note": "factor recovery consistent across studies and forms"}
• {"subgroup": "Personal accomplishment / professional efficacy", "grade": "Low", "note": "separability contested; bifactor reanalysis suggests a dominant general factor"}

Convergent and discriminant validity Moderatewell-establishedevidence form: mixed

indirect predominantly non-UK occupational and healthcare samples

MBI-GS subscales correlate in the expected directions with job satisfaction, work engagement and psychological distress in large workplace samples (Bravo D. 2021). Emotional exhaustion is the dimension most strongly tied to health and work-strain criteria, whereas personal accomplishment correlates weakly and sometimes inconsistently with the other two subscales, part of the same evidence that drives the structural debate (Aguayo-Estremera R. 2024). Convergent evidence against alternative burnout instruments is documented in the comparative literature (for example moderate-to-high correlations with the Oldenburg Burnout Inventory; Demerouti E. 2003).

Criterion validity: reference standard Very lowcontestedevidence form: canonical

indirect the absence of a diagnostic standard is generic, and the critical literature draws on mixed occupational samples

There is no diagnostic reference standard for burnout, so the MBI cannot be validated against a gold-standard diagnosis; the manual provides no clinical cut-off, and a long-standing cautionary analysis showed the proposed cut-off points lacked cross-national and clinical validity (Schaufeli W. 1995). A recent critique argues more strongly that the MBI does not measure the construct it claims to and that burnout itself is poorly defined (Bianchi R. 2024). A COSMIN-based systematic review of occupational burnout measures graded the overall evidence and highlighted persistent gaps in criterion validity against health standards (Shoman Y. 2021).

Criterion validity: organisational Moderatewell-establishedevidence form: canonical

indirect nursing and teaching cohorts outside the UK, outcomes often intention-to-leave rather than recorded organisational data

A two-year prospective cohort of 509 teachers found high depersonalisation predicted long-term (>=30 day) sickness absence after full adjustment (relative risk 1.80, 95% CI 1.05 to 3.09), while emotional exhaustion lost significance after adjustment and low personal accomplishment was unrelated (Salvagioni D. 2022). Burnout measured by the MBI is associated with turnover intention in nurses (Zheng J. 2024) and with intention to leave the profession in a national hospital-nurse study (Bruyneel A. 2023), and a systematic review links nurse burnout to patient safety and quality-of-care outcomes (Li L. 2024). Associations are consistent in direction but vary by subscale and are mostly self-reported or intention-based rather than registered absence or turnover.

Internal consistency Highwell-establishedevidence form: mixed

indirect pooled across many non-UK occupational samples

A reliability generalisation meta-analysis of 84 studies found mean subscale alphas generally between 0.70 and 0.80, with personal accomplishment and depersonalisation frequently below levels recommended for high-stakes individual decisions (Wheeler D. 2011); an updated meta-analysis reported average alphas of 0.71 to 0.88 across dimensions (Aguayo-Estremera R. 2024). Individual studies commonly report emotional exhaustion as the most reliable subscale (for example MBI-GS alphas 0.72 to 0.86 in Colombian workers, Bravo D. 2021; MBI-GS9 alpha and omega 0.84 to 0.91, a derivative short form, Wang A. 2024). Depersonalisation and personal accomplishment are the weakest-performing subscales for internal consistency.

Sub-grades (evidence differs by subgroup):

• {"subgroup": "Emotional exhaustion", "grade": "High", "note": "alpha typically 0.80 to 0.90"}
• {"subgroup": "Depersonalisation & personal accomplishment", "grade": "Moderate", "note": "means often 0.70 or below; unreliable for individual classification"}

Test-retest reliability Lowthin

indirect the one retrievable coefficient set is item-level ICCs from an Iranian medical-personnel sample, not a UK working-adult subscale-level retest, and the retest interval was not stated in the retrieved text

Test-retest reliability is under-reported for the fielded MBI. The MBI manual (Mind Garden, behind the commercial licence and not retrievable in the latest review pass) reports stability from one month to one year, as cited secondarily by the Persian MBI-HSS-MP study. The only numeric coefficients extractable in the latest review pass were item-level ICCs of 0.70 to 0.92 from that Persian medical-personnel validation; subscale-level retest coefficients in UK working adults were not located. This is recorded as thin rather than well-established.

Measurement invariance Moderatecontestedevidence form: mixed

indirect US physician, Colombian worker and teacher samples, no dedicated UK invariance study located

An item-response-theory differential item functioning analysis of 6,577 US physicians detected statistically significant DIF across age, gender and specialty in nearly every MBI item, but the impact on subscale scores was small (mean differences <0.10 SD), so measured group differences are not largely artefacts of item bias (Brady K. 2021). Full metric and scalar invariance of the MBI-GS was supported across gender, age and socioeconomic status in Colombian workers (Bravo D. 2021), and cross-gender factorial invariance was broadly supported in teachers with some items flagged (Byrne B. 1994). Invariance across occupations and languages is less thoroughly established than across sex and age.

Sub-grades (evidence differs by subgroup):

• {"subgroup": "Across sex and age", "grade": "Moderate", "note": "metric and scalar invariance supported in several samples"}
• {"subgroup": "Across occupation / specialty", "grade": "Low", "note": "significant item-level DIF detected, though small in aggregate impact"}

Responsiveness and MIC Lowthinevidence form: canonical

indirect intervention samples are largely non-UK physicians; no MIC anchor established

A systematic review and meta-analysis of interventions found that structured programmes reduced emotional exhaustion and overall burnout scores in physicians, evidence that MBI scores move with change (West C. 201631279-x)). However, no established minimal important change (MIC) value for any MBI subscale was located, and the COSMIN review noted responsiveness as an under-evidenced property (Shoman Y. 2021).