Skip to main content
v2026.11,986 entries · CC-BY 4.0

Copenhagen Burnout Inventory: Scoring and Examples

Score the Copenhagen Burnout Inventory with checked examples, missing-response rules, reverse coding and reproducible Excel, R and Python calculations.

Written and maintained by CASRAI Editorial Board

Published

On this page

The Copenhagen Burnout Inventory (CBI) is a 19-item questionnaire covering personal, work-related and client-related exhaustion. For research scoring, calculate each scale separately on a 0–100 metric, reverse the positively worded work item and check the number of answered items before taking an average. The examples below show how those checks change the result.

Copenhagen Burnout Inventory definition and scope

The CBI distinguishes general physical and psychological exhaustion from exhaustion that respondents attribute to work or to working with clients. Its personal scale has six items, its work scale seven and its client scale six. The National Research Centre for the Working Environment (NFA) distributes the English questionnaire and its scoring instructions. Use those named scales in a research dataset so a reader can identify which outcome was measured.

NFA permits non-commercial and commercial use when the specified PUMA study publication is clearly cited: Borritz and colleagues, Burnout among employees in human service work: design and baseline findings of the PUMA study, Scandinavian Journal of Public Health, 2006, 34, 49–58. Its official CBI page supplies the downloads and this condition.

CBI scoring formula and response coding

The NFA English scoring sheet assigns ordinary frequency responses 100 for always, 75 for often, 50 for sometimes, 25 for seldom and 0 for never/almost never. Degree responses follow the corresponding descending sequence from very high to very low degree. Work item 7, about energy for family and friends, has the opposite direction.

S = (x₁ + x₂ + … + xₖ) / k. Here S is one respondent’s score on one scale, xᵢ is the coded value of answered item i after any reversal, and k is the number of answered items in that scale. An unanswered item contributes neither a value nor a place in the denominator. On the 0–100 coding used here, the reverse transformation is xᵣ = 100 − x, where x is the ordinary code and xᵣ is the reversed code.

These equations are an implementation of an item average. Keep the item-level values and the resulting scale score in separate columns. A survey export using labels or another numeric convention needs an explicit mapping before this calculation; a column containing the number 4 does not tell another analyst which answer the respondent selected. The distinction between ordered responses and their numerical coding also matters in Likert scale analysis.

Worked example: reverse coding and missing responses

Illustrative composite: all response values in this section are invented for arithmetic demonstration, not observations from a study. Suppose the six personal item codes are 75, 50, 50, 25, 75 and 0. Their sum is 275, so the personal score is 275 / 6 = 45.833333, displayed as 45.8. The zero is an answered response and remains in the denominator.

Now take seven ordinary work-item codes in the NFA sheet’s item order: 75, 50, 25, 75, 50, 25 and 75. The final value represents frequent energy for family and friends. Reverse that value to 100 − 75 = 25. The scoring vector becomes 75, 50, 25, 75, 50, 25 and 25. Its sum is 325; the work score is 325 / 7 = 46.428571, displayed as 46.4.

Skipping reversal would give 375 / 7 = 53.571429. In this example, a coding error therefore adds 7.142857 points. By contrast, a midpoint response of 50 remains 50 after reversal, so testing only midpoint responses would fail to expose the error. Include an asymmetric example like this when checking a scoring script.

For a missing-response check, retain only work items 1, 2, 3 and 7 from that example. Their scored values are 75, 50, 25 and 25: 175 / 4 = 43.75. Under NFA’s rule, four answered work items are enough to score. If only the first three remain, return a missing scale score. Do not output 150 / 3 = 50 for that record, even though a generic averaging function can calculate it.

Replacing those three unanswered items with zeros would instead produce 175 / 7 = 25. That changes the meaning of the data: the calculation now treats nonresponses as low-exhaustion answers. Preserve the distinction in the exported data and in the scoring function. The worked examples deliberately retain full precision until the displayed result is rounded.

Interpreting CBI scores and reference values

A CBI score identifies the location of an item average on its coded scale. The NFA reference document reports historical PUMA means separately for personal, work and client burnout. It also describes a grouping at 50 or more for that study. Carry that study context with the number; do not turn its grouping into an individual diagnosis.

Result or comparison Reading guidance Basis
Individual work score of 46.4 in the example A mean of coded work responses; it is not 46.4% of people with burnout. NFA item-average rule and the calculation above.
Comparison with a published mean Name the scale and reference population alongside the value. NFA reports separate scale and population results.
Grouping at 50 or more Describe it as the threshold used in the cited PUMA summary. NFA reference document, February 2004, page 8.

The downloadable data table below joins scoring checks to the historical PUMA reference means and their scale-specific sample sizes. It is a crosswalk, not a new norming study. For a comparison, write the reference document’s date and population into the analysis notes. A score from a different workforce or survey round needs its own context rather than a label inferred solely from a historical mean.

Keep individual item averaging separate from summarising a sample of people. For example, three illustrative participant work scores of 25, 50 and 75 have a sample mean of 50 and a sample standard deviation of 25. They do not establish the distribution in any actual workforce. Report the number of scored participants with these descriptive statistics, rather than using an item count as the sample size.

Assumptions and common scoring mistakes

The scoring assumptions are concrete: the columns represent the intended items, numeric codes reflect the stated answer categories, reversal has happened once, and the answered-item requirement has been met. Treat these as checks on the implementation. A successful spreadsheet formula establishes none of them by itself.

Keep a short item map containing the scale, item identifier, answer mapping and reversal flag. NFA recommends mixing the questionnaire items rather than displaying them in the scoring sheet’s order. Consequently, the last column in a custom survey export is not automatically work item 7. Match the energy item by identity before applying the transformation.

Distinguish a missing answer inside an administered scale from a scale that was not administered. If a questionnaire omitted client questions, a zero client score would misrepresent that design choice. Record what was administered and keep its outcome separate from the other scales. For item nonresponse, document the scoring rule alongside the broader missing-data analysis.

APA’s JARS–Quant reporting standards ask researchers to describe instrument quality and report score reliability for the analysed sample. Report your own scale-level reliability evidence where appropriate; historical coefficients do not describe a new dataset automatically. The Cronbach’s alpha guide explains a common reliability statistic and its interpretation.

How to report CBI results: an APA example

Identify the version, scales administered, response coding, reversal rule and missing-response handling in the methods. In the results, report a participant count and summary statistics for each analysed scale. These choices make it possible to reconstruct the outcome and are consistent with APA’s JARS–Quant Table 1 recommendations on measures, missing data and descriptive results.

Illustrative composite, not a real study: “Work-related burnout was scored using the seven-item CBI work scale on a 0–100 metric. The energy item was reverse coded, and scores required at least four answered items. Across three illustrative participants, work-related burnout scores were M = 50.0, SD = 25.0.”

The example uses italic statistical letters, spaces around equals signs and one decimal for the displayed mean and standard deviation, following the APA 7th edition Numbers and Statistics Guide. In a manuscript, replace the example’s disclosure and numbers with the actual design and results. Retain a reproducible calculation at full precision, then choose display precision consistently. Do not imply that a descriptive result demonstrates an intervention effect.

Calculate CBI scores in Excel, R and Python

These equivalent examples implement the work scale using the seven ordinary 0–100 codes from the worked example. Inputs must already be valid numeric response codes; blank, NA or None denotes an unanswered item. Each implementation reverses the energy response once and checks the four-answer minimum. Keep raw and scored values distinguishable when applying the logic to a whole dataset.

Before running a reusable scoring routine, check that every nonmissing input belongs to the allowed set: 0, 25, 50, 75 or 100. A sentinel such as 999 must be converted to a missing value before averaging. Save both the number of answered items and the resulting score, so an analyst can distinguish an excluded record from a valid low score without reconstructing the entire calculation.

Excel

Place the ordinary codes in B2:H2, with work item 7 in H2. In I2, calculate its reversed value; in J2, calculate the score. Keep missing inputs genuinely blank. Microsoft documents that AVERAGE ignores blank cells while including numeric zeros.

I2: =IF(H2="","",100-H2)
J2: =IF(COUNT(B2:G2,I2)>=4,AVERAGE(B2:G2,I2),"")

R

The R base mean function supports removal of NA values. The answer-count guard is still necessary: removing missing entries does not enforce an instrument’s scoring threshold.

x = c(75, 50, 25, 75, 50, 25, 75)
if (!is.na(x[7])) x[7] = 100 - x[7]
score = if (sum(!is.na(x)) >= 4) mean(x, na.rm = TRUE) else NA_real_
score

Python

Python’s statistics.mean provides the arithmetic mean. This example filters None explicitly before averaging and leaves the raw vector intact.

from statistics import mean
raw = [75, 50, 25, 75, 50, 25, 75]
scored = raw.copy()
if scored[6] is not None:
    scored[6] = 100 - scored[6]
answered = [x for x in scored if x is not None]
score = mean(answered) if len(answered) >= 4 else None
print(score)

The complete example returns approximately 46.428571 in each implementation. Replacing work items 4–6 with missing values should produce 43.75; then removing item 7 should produce a missing score. Finally, test a complete row of ordinary zeros: reversal makes the energy value 100, so its work average is 100 / 7. This catches accidental removal of valid zero responses.

Sources

Data: CBI scoring requirements and historical PUMA reference values

Original crosswalk joining NFA English CBI scoring requirements to the scale-specific means and sample sizes in the February 2004 NFA PUMA summary. These historical human-service workforce data are descriptive reference figures, not universal diagnostic cutoffs. Columns identify which source supplies each value.

Period covered: Historical reference summary published February 2004. Coverage: Denmark.

CBI scoring requirements and historical PUMA reference values
ScaleItems (items)Minimum answered (items)Reverse-coded itemHistorical PUMA mean (points (0–100))Historical scored sample (respondents)
Personal burnout63None35.91898
Work-related burnout74Item 7: energy for family and friends331910
Client-related burnout63None30.91752

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Copenhagen Burnout Inventory: Scoring and Examples

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →