Kessler K-10: administration and interpretation in practice

Two errors follow the K-10 around. It gets presented as a depression or an anxiety scale, and it is neither: it measures non-specific psychological distress, a number that does not tell you what produces it. And there are two scoring conventions — 1 to 5 for a 10–50 range, and 0 to 4 for a 0–40 range — with much material pairing one with the other’s interpretation table. Score on 0-to-4 anchors, then read the Australian band table, and you classify every client ten points low: a “22, high” becomes a “12, low.”

What the K-10 measures and what backs it

The K-10 measures global psychosocial distress with questions about nervousness, agitation, psychological fatigue and depressed mood, as described by the Australian Bureau of Statistics (ABS), author of the most complete official document on the instrument. Kessler and Mroczek built it by refining 45 initial questions, and the published scales were constructed with item response theory for the redesign of the US National Health Interview Survey: they wanted a brief measure of non-specific distress, not a disorder instrument. The ABS fixes its scope:

“The K10 instrument is not a diagnostic tool, but an indicator of current psychological distress, where very high levels of distress may signify a need for professional help.”

The PHQ-9 maps criteria for a named disorder and the GAD-7 does the same for generalized anxiety; the K-10 maps criteria for nothing. If your question is “depression or anxiety?”, it is the wrong instrument by design: administer the other two.

It was developed in 1992, published in 1994 and revised in 2001; its construction is in Kessler et al. (2002). The study customarily cited as a condition of use is a different one, Kessler et al. (2003), which validated the scales against serious mental illness in an enriched sample of 155 people, with an AUC of 0.854: real evidence, and also a small sample. Against 30-day CIDI/DSM-IV mood and anxiety disorders in the Australian national survey it performs better: AUC 0.90 in Furukawa and colleagues (2003), a third study.

Structure, versions, and the two scoring conventions

Ten items, five frequency levels, and two versions the ABS documents separately: the 1994 version asks about “the past 4 weeks” and uses “sad”; the 2001 version, about “the past 30 days,” uses “depressed” and adds a third conditional follow-on. Always record the version. No completion time here: the “2 to 5 minutes” figure circulates everywhere and we could not locate its source.

Convention	Value per item	Total range	Who uses it
Australian	1 to 5	10 to 50	ABS and Australian surveys; the Mexican validation
United States	0 to 4	0 to 40	Typical US use; current normative studies

Neither is “the correct” one: the seven-country European normative study in Scientific Reports (2023) reports everything on the 0–40 metric. Two rules prevent the ten-point error. Identify the convention from the form, not the table: if the printed anchors read 0, 1, 2, 3, 4, your total runs 0 to 40 and the 10–50 bands do not apply. And write it into the note: “K-10: 22 (1–5 scale, 10–50 range).”

Where the severity bands actually come from

The ABS says it plainly and almost nobody repeats it: no universally agreed categories or groupings exist for K-10 scores. What circulates as “the K-10 cutoffs” is several Australian systems, all on the 10–50 convention.

Score (10–50)	ABS grouping	CRUfAD and GPcare (primary care)
10–15	Low	10–19: likely to be well
16–21	Moderate	(same range as above)
22–29	High	20–24 mild; 25–29 moderate
30–50	Very high	30–50 severe

The ABS is explicit that the CRUfAD and GPcare groupings are used to monitor distress rather than to identify the presence of a disorder; specialist services use the same boundaries worded conditionally and a third, “Plain English” system uses three levels. That ABS paper is archived: we did not verify which system is in force in Australia today.

The Mexican evidence: one validation we could locate

The only Mexican validation we located is Vargas Terrez and colleagues (2011), in Salud Mental: 280 adults at two primary-care centres in Coyoacán, Mexico City, in 2010, with the MINI as the gold standard. It is good evidence.

Item	Value
Internal consistency	Cronbach’s alpha of 0.901, single factor
Cut point for depression	21 — sensitivity 78.7%, specificity 79.0%, AUC 87.0%
Cut point for anxiety	22 — sensitivity 72.4%, specificity 73.8%, AUC 82.4%
Positive / negative predictive value, depression	57.8% / 91.0%
Positive / negative predictive value, anxiety	24.7% / 95.7%

Those optimal cut points do not line up with the Australian band boundaries: 21 is the ceiling of “moderate” and 22 the floor of “high.” The authors conclude it is effective for depression and that for anxiety “the efficiency is weak.” Positive predictive value for anxiety was 24.7%, against a negative predictive value of 95.7%. A low K-10 helps you rule out; a high one is an invitation to interview, not a conclusion.

We located no Mexican normative data either: no percentiles or means to offer you.

Other Spanish-language evidence, and why it is not enough

Translation is not validation, and reliability is not diagnostic accuracy: a high alpha says the items measure something in common, not that the score finds cases.

Spain. Sánchez-Moreno and colleagues (2025), in 887 people aged 65 and over: a confirmed one-factor structure, an alpha of 0.944 and convergent validity against the GDS-5. No diagnostic-accuracy analysis.
Ecuador. Larzabal-Fernández and colleagues (2024), with 5,132 adolescents: the ten-item version was not the best-fitting one and they prefer the nine-item version there.
Colombia, Peru, Chile and Argentina. We located no published K-10 validation in any of those countries.

The review by Stolk and colleagues (2014) carries the most weight: evidence for the validity of the original K-6 and K-10 in culturally diverse populations is limited, questions arise about the conceptual and linguistic equivalence of translated versions, and clinical norms are lacking — hence their call for caution. And in that European study, on the same 0–40 metric, the optimal threshold against self-reported mental disorder ranged from 7.5 (Italy) to 16.5 (Spain). Never change a cut point on your own authority, and do not treat the one you use as universal.

The K-6 is not the K-10

The K-6 is the K-10 minus four questions, and the ABS names the cost: truncated, it cannot rely entirely on the validation of the instrument it came from. Its range is 0 to 24 or 6 to 30 depending on the convention, and the standard rule classifies scores of 13 to 24 as probable serious mental illness — a threshold of the K-6 and only the K-6. It is the K-6, not the K-10, that sits in Mexican national surveillance. A study of perinatal distress in Mexican fathers in Salud Mental records that the 2016 ENCODAT survey administered it in 56,877 interviews.

What a K-10 score does not tell you

It does not name a disorder, and it does not diagnose. It is engineered not to distinguish, and the ABS says so expressly. A 30 is not “severe depression”: it is high distress of undetermined origin.
It performs unevenly near the floor. In the European study, 9.4% scored the minimum on the 0–40 metric and only 0.1% the maximum: very low scorers have little room to show change.
It does not ask about risk. The K-10 contains no suicidal ideation item, and that does not delegate the topic. Any risk disclosure calls for an immediate clinical risk assessment under your own protocol and training, documented in the record. This guide will not hand you that protocol — it comes from your training and the applicable professional guidelines — and make sure the client leaves knowing the crisis line serving their location, plus local emergency services.

Where the K-10 earns its place is in repetition, as in measuring therapy progress: same instrument, same version, same convention, same moment of administration.

Licensing: free of charge, but not public domain

The ABS, reporting Andrews and Slade, contrasts the K-10 with the GHQ: the latter is proprietary and is paid for each time it is used, whereas the K-10 “is in the public domain and may be used without charge.” Keep the half that holds: you can administer it without buying anything from a distributor, rare in this catalogue — the contrast is in the guide on test licensing. But “public domain” would also mean no copyright and no duty to attribute, which collides with the condition usually attached to it: acknowledging a WHO copyright. Treat it as free to use with mandatory attribution: do not translate or redistribute it on your own.

The formal terms come not from the ABS but from the rights holder’s canonical page: the National Comorbidity Survey page at Harvard Medical School, hcp.med.harvard.edu/ncs/k6_scales.php. While this guide was being prepared that page did not respond — a server error behind a bot filter — so we could not quote it verbatim. What is customarily summarized from it is two conditions: no permission request is needed, and use requires citing the 2003 study and acknowledging a World Health Organization copyright. We do not present those as verified: open the page before reproducing the instrument. Nor did we locate that WHO acknowledgement’s exact wording published anywhere, so the half you can meet today is the citation; put it at the foot of the form and in any report where you state the score:

Kessler RC, Barker PR, Colpe LJ, et al. Screening for serious mental illness in the general population. Archives of General Psychiatry. 2003;60(2):184-189. doi:10.1001/archpsyc.60.2.184

One gap we cannot close: we located no identifiable official Spanish edition of the K-10. The Mexican paper records only that the scale “has been translated into Spanish,” naming no translator, publisher or version, and we found no equivalent to the Spanish-for-Mexico PDF Pfizer publishes for the PHQ-9. The files in circulation carry no attribution and no date: if you use one, do not present it as official.

Where the score lives in the note

A score is a relevant result from a diagnostic aid and belongs in the progress note. Many jurisdictions regulate what that note must contain; in Mexico, for example, NOM-004-SSA3-2012 requires it at numeral 6.2.3 and adapts those notes to the nature of a psychological service at 5.16 — see the guide to the Mexican clinical record standard. Verify the rules where you practice; this is documentation guidance, not legal advice.

Record six data points: instrument and version, scoring convention, raw score, band plus the system it came from, mode of administration, and date. Then your clinical reading and what changed in the plan. In a SOAP note the score goes in the O and its interpretation in the A. Recording the band alone leaves a gap nobody can reconstruct, and heads the list of clinical documentation mistakes.

“K-10, 2001 version, self-administered. Score 24 (1–5 scale, 10–50 range); high band per ABS groupings. Previous baseline 31. The instrument does not separate anxious from depressive components: adding PHQ-9 and GAD-7.”

Scoring without slipping ten points and comparing against baseline is easy with one client and heavy with thirty. That is where a platform helps: gesell.ai includes built-in validated scales with automatic scoring — the PHQ-9 and GAD-7 among them — and keeps the structured chart where today’s score sits next to the previous ones; anything you administer outside it records just as well if you write those six data points. The scale contributes the data point as decision support you evaluate; the clinical judgment remains yours.