Harvey–Bradshaw Index Calculator
Harvey–Bradshaw Index Calculator
Five items scored on yesterday alone — the simplified CDAI. It correlates with the CDAI at r = 0.93, and regression puts one HBI point at about 27 CDAI points, which is how to read a trial endpoint against a clinic score.
Harvey–Bradshaw index
5 items → HBIWell-being poor (2), mild abdominal pain (1), four liquid stools yesterday (4), no abdominal mass (0), no complications (0)
Scoring
under 5 remission · 5–7 mild · 8–16 moderate · 17 and above severe
CDAI ≈ 27 × HBI (Best 2006, 224 patients)
- yesterday, not this week
- every item is scored on the previous day. That is the whole point of the index and the reason it can be done in a clinic without a diary, and it is also why a single HBI is noisier than a CDAI built from seven days
- the stool item is uncapped
- one point per liquid or soft stool, with no ceiling. Four of the five items together can reach only 18; the stool count alone can exceed that, so in a patient with frequent diarrhoea this is effectively a stool-frequency score
- 27 CDAI points per HBI point
- Best’s 2006 regression over single visits of 224 patients, pooling the development data of both indices. It is a group relationship with wide scatter, not a conversion: use it to orient a clinic score against a trial endpoint, never to report a CDAI you did not measure
- r = 0.93, and then r = 0.80
- Harvey and Bradshaw’s own 112 patients, scored on both indices at the same visit, gave a correlation of 0.93. Pooled across the PRECiSE 1 and 2 trials it was 0.800 (95% CI 0.780 to 0.819). The weaker figure is the one that describes a trial population
- four levels of abdominal mass
- 0 none, 1 dubious, 2 definite, 3 definite and tender. The fourth level is routinely dropped in reproductions, which silently caps the item
Worked example
Well-being poor (2), mild abdominal pain (1), four liquid stools yesterday (4), no abdominal mass (0), no complications (0)
2 + 1 + 4 + 0 + 0 = 7 points
7 sits at the top of the 5–7 mild band — one more liquid stool would move it to moderate
By the Best 2006 regression, 7 HBI points is of the order of 190 CDAI points, which is inside the CDAI's 150–220 mild band but below the 220 floor most trials use for entry
That orientation is the useful part. It does not mean this patient has a CDAI of 190: the relationship is a group regression with wide scatter and the pooled trial correlation is 0.80, not 0.93
Note where the score came from — four of the seven points are the stool count, which is the one uncapped item in the index
The five items
| Item | Levels | Maximum |
|---|---|---|
| General well-being yesterday | 0 very well, 1 slightly below par, 2 poor, 3 very poor, 4 terrible | 4 |
| Abdominal pain yesterday | 0 none, 1 mild, 2 moderate, 3 severe | 3 |
| Liquid or soft stools yesterday | 1 point each | no ceiling |
| Abdominal mass | 0 none, 1 dubious, 2 definite, 3 definite and tender | 3 |
| Complications | 1 point each: arthralgia, uveitis, erythema nodosum, aphthous ulcers, pyoderma gangrenosum, anal fissure, new fistula, abscess | 8 |
How the HBI and the CDAI line up, and where they do not
| Figure | Cohort | What it says |
|---|---|---|
| r = 0.93 | 112 patients scored on both indices at the same visit (Harvey and Bradshaw, 1980) | The original claim that the HBI carries essentially the same information as the CDAI. |
| r = 0.800 (95% CI 0.780–0.819) | Pooled across the PRECiSE 1 and 2 certolizumab trials | Weaker in a trial population. About a third of the variance in one index is not explained by the other. |
| 27 CDAI points per HBI point | Single visits of 224 patients, pooling both indices’ development data (Best, 2006) | A group regression slope. HBI 5 maps to about 135 and HBI 8 to about 216 — close to, but not the same as, the CDAI’s own 150 and 220. |
| Remission: HBI 4 or less | ECCO-ESGAR 2018, Table 5 | Matches the four-band table’s “under 5”. |
| HBI 3 or less implies CDAI remission | ECCO-ESGAR 2018 | The figure to use when the two indices have to agree. 4 is remission on one and not reliably on the other. |
The index that exists because the CDAI is impractical
The CDAI needs a seven-day patient diary, a haematocrit and a standard body weight. Harvey and Bradshaw’s answer, published as a single page in the Lancet in 1980, was to keep five items, score the first three on yesterday alone, and add them up without weights. It can be done in a clinic room in under a minute, which is why most real-world Crohn’s cohorts are scored with it and most trials are not.
The relationship between the two indices is the content of this page. In Harvey and Bradshaw’s own 112 patients, scored on both instruments at the same visit, the correlation was 0.93. Pooled across the PRECiSE 1 and 2 trials it was 0.800, with a 95% confidence interval of 0.780 to 0.819. Best’s 2006 regression, over single visits of 224 patients pooling the development data of both indices, found that each HBI point corresponded to about 27 CDAI points. Those three numbers do different jobs. The 0.93 says the HBI captures the same construct. The 0.80 says that in a trial population roughly a third of the variance in one index is not explained by the other. The 27 lets you orient a clinic HBI against a published CDAI endpoint — HBI 5 is of the order of 135 CDAI points and HBI 8 of the order of 216, close to but not the same as the CDAI’s own 150 and 220.
What that means in practice: a trial reporting “clinical remission, CDAI under 150” and a clinic recording “HBI 4, in remission” are making compatible but not interchangeable statements, and the 27-point slope is a group relationship with wide scatter rather than a conversion you can quote for one patient. ECCO-ESGAR’s stricter observation is the useful one when the two have to agree: an HBI of 3 or less is very likely to be CDAI remission too.
Two scoring traps. The stool item is one point per liquid stool with no ceiling, so in frequent diarrhoea it dominates everything else — eight stools is already the moderate band before any other item is scored. And the abdominal mass item has four levels, not three: “definite and tender” scores 3, and reproductions that stop at “definite” silently cap it. Finally, the two published band tables disagree above 8: the four-band table calls 8 to 16 moderate while ECCO-ESGAR calls 8 to 9 and above severe, so say which you are using. This index scores symptoms, not inflammation. It can be high with a normal mucosa — bile-acid diarrhoea, a stricture, bacterial overgrowth, coexisting irritable bowel syndrome — and low with active endoscopic disease. That mismatch is why treat-to-target moved off symptom indices and onto objective markers. The faecal calprotectin interpreter and the CRP unit converter are what to check before reading a high score as inflammation, and the Rutgeerts score is the instrument for recurrence after a resection. A score is not a diagnosis and a figure from a cohort is not a probability for one patient. This states what the number meant in a named study; the clinician in front of the patient decides what follows. Thresholds here are the published ones; reference intervals and assay units are method- and laboratory-dependent and your own laboratory’s interval takes precedence.
Frequently asked questions
What is a normal Harvey–Bradshaw index?
Under 5 is remission on the conventional four-band table, and ECCO-ESGAR gives the same boundary as 4 or less. 5 to 7 is mild and 8 to 16 moderate on that table, with 17 and above severe. ECCO-ESGAR disagrees above 8, calling 8 to 9 and above severe, so a score of 12 is moderate under one source and severe under the other.
How do I convert an HBI to a CDAI?
You cannot, for an individual. Best’s 2006 regression over 224 patients gives about 27 CDAI points per HBI point, and this page shows that product beside the score so a clinic HBI can be oriented against a trial’s CDAI endpoint. But it is a group slope with wide scatter, and the pooled trial correlation between the two indices is 0.80 rather than 0.93 — so the product is an orientation, not a CDAI you can report.
Is the stool count really uncapped?
Yes. One point per liquid or soft stool yesterday, with no ceiling, which is why the index has no real maximum. The other four items together can reach only 18. In a patient with ten liquid stools a day the HBI is largely a stool-frequency score, and that is worth saying when the other items are unremarkable.
Which day does the index refer to?
Yesterday. All three symptom items are scored on the previous day rather than averaged over a week, which is deliberate — it reduces recall bias and is what makes the index practical. The cost is precision: a single bad day moves an HBI in a way it could not move a seven-day CDAI.
Does a low HBI mean the bowel has healed?
No. The index scores how a patient felt yesterday, an abdominal examination and a count of extraintestinal manifestations. None of those measures mucosal inflammation, and the mismatch runs both ways — a normal mucosa with bile-acid diarrhoea or a stricture scores high, and active endoscopic disease can score low. Faecal calprotectin, CRP and endoscopy are what close that gap.
Related calculators
References
- New Zealand Society of Gastroenterology. Harvey-Bradshaw Index patient brochure. nzsg.org.nz.
- Den norske legeforening. Harvey-Bradshaw indeks scoring form. nettkurs.legeforeningen.no.
- Best WR. Predicting the Crohn’s disease activity index from the Harvey-Bradshaw Index. Inflamm Bowel Dis. 2006;12(4):304–10.
- Sturm A, Maaser C, Calabrese E, et al. ECCO-ESGAR guideline for diagnostic assessment in IBD part 2: IBD scores and general principles and technical aspects. J Crohns Colitis. 2019;13(3):273–84. doi:10.1093/ecco-jcc/jjy114
- Egyptian Journal of Internal Medicine. Clinical and endoscopic characteristics of inflammatory bowel disease: a single-centre study. Egypt J Intern Med. 2022;34:115, Table 1.
Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/