Cervical Length Preterm Birth Risk Interpreter
Cervical Length Preterm Birth Risk Interpreter
The published transvaginal thresholds, read against the obstetric history – because history moves the predictive value far more than the millimetres do: 81% with a prior spontaneous preterm birth against 26 to 39% in the general population.
Published thresholds by history
Transvaginal measurement onlyCervical length 22 mm, 20 completed weeks, singleton with no prior spontaneous preterm birth, measured transvaginally
The thresholds, and why history matters more than millimetres
Lower published thresholds: 20 mm, 15 mm, 10 mm · window 16 to 24 weeks, anatomy scan 18 to 22 weeks
- 25 mm or less
- SMFM’s Consult Series defines a midtrimester short cervix as 25 mm or less, as a graded recommendation, and notes that all the definitions apply only where there is no cervical dilation. Exactly 25 mm IS short on that wording; 25.1 mm is not
- transvaginal, and only transvaginal
- the Consult Series’ first recommendation is that any cervical length used to guide treatment must be measured transvaginally. Transabdominal assessment is acceptable as a screen at the anatomy scan. The thresholds were not derived transabdominally and this page refuses to apply them to that route
- history changes the test, not the number
- the positive predictive value of a short cervix for preterm birth before 34 to 35 weeks reaches 81.0% with a prior spontaneous preterm birth and is 26.3 to 39.1% in the general population. Two to three times the predictive value from identical millimetres, which is why this is an interpreter over history rather than a threshold check
- high specificity, low sensitivity
- in nulliparas measured at 16 to 22 weeks, 25 mm gave sensitivity 8.0%, specificity 97.8% and positive predictive value 16.2% for birth before 37 weeks, and 20 mm gave 4.1%, 98.8% and 15.5%. For birth before 32 weeks at 25 mm: 23.9%, 97.7% and 7.4%. Tightening the threshold raises specificity and lowers sensitivity, as any threshold must
- the extremes carry the events
- Heath and colleagues found a length under 15 mm in 1.7% of an unselected population at 22 to 24 weeks, and those patients accounted for 86% of preterm births before 28 weeks and 58% before 32 weeks. Where no length is measurable the Consult Series reports a 75% risk of delivery before 32 weeks, with a median diagnosis to delivery interval of 3 weeks – and that not all such patients deliver before 32 weeks
- twins have no agreed threshold
- the Consult Series says in terms that in twin pregnancies there is no consensus on the measurement at which to diagnose a short cervix, and recommends against routine intervention outside clinical trials. The singleton numbers are not transferable
- no centile here, and why
- cervical-length centiles for gestational age are a published table and this engine cannot interpolate one. The 10th centile is quoted as about 26 mm at 18 to 24 weeks, which is a useful single figure, but a fitted curve would misclassify at the extremes, which is where the decision sits. Only the discrete published thresholds are applied
Worked example
Cervical length 22 mm, 20 completed weeks, singleton with no prior spontaneous preterm birth, measured transvaginally
Transvaginal, singleton without a prior spontaneous preterm birth, and 20 weeks is inside the 16-to-24-week window, so the published thresholds apply
22 mm is at or below 25 mm, so it is a short cervix on SMFM's definition; it is above 20 mm, so it does not reach the lower threshold
Set 25.0 mm and it is still short, because the definition is 25 mm OR LESS; set 25.1 mm and it is not
Set 20.0 mm and the result moves to the lower threshold band; set 9.9 mm and it moves below ten millimetres
Change the history to a prior spontaneous preterm birth and the page reports that instead, because the predictive value changes from 26 to 39% to 81% on the same millimetres
Change the route to transabdominal and the page refuses the thresholds; change the gestational age to 28 weeks and it says the measurement is outside the validated window
The published thresholds and what sits at each
| Length | What the source says | Source |
|---|---|---|
| 25 mm or less | Midtrimester short cervix, in the absence of cervical dilation; about a six-fold risk increase at 24 weeks in an unselected cohort | SMFM Consult Series |
| 21 to 25 mm | A separate graded band in the Consult Series | SMFM Consult Series |
| 20 mm or less | The band carrying the Consult Series’ strongest graded recommendation, diagnosed before 24 weeks | SMFM Consult Series |
| Under 15 mm | 1.7% of an unselected population at 22 to 24 weeks, accounting for 86% of births before 28 weeks and 58% before 32 | Heath and colleagues, via the Consult Series |
| 10 to 25 mm, no dilation | A band the Consult Series addresses as a whole | SMFM Consult Series |
| No measurable length | 75% risk of delivery before 32 weeks, median 3 weeks from diagnosis; not all such patients deliver before 32 weeks | SMFM Consult Series |
Why history matters more than the millimetres
| Population and threshold | Sensitivity and specificity | Positive predictive value |
|---|---|---|
| Nulliparas, 25 mm, birth before 37 weeks | 8.0% and 97.8% | 16.2% |
| Nulliparas, 20 mm, birth before 37 weeks | 4.1% and 98.8% | 15.5% |
| Nulliparas, 25 mm, birth before 32 weeks | 23.9% and 97.7% | 7.4% |
| Nulliparas, 20 mm, birth before 32 weeks | 14.9% and 98.8% | 8.6% |
| Nulliparas, 20 mm measured at 22 to 30 weeks | Not stated | 20.8% before 37 weeks, 3.9% before 32 weeks |
| With a prior spontaneous preterm birth, birth before 34 to 35 weeks | Not stated | 81.0% |
| General population, birth before 34 to 35 weeks | Not stated | 26.3 to 39.1% |
Why the same millimetres mean different things
A midtrimester cervical length of 25 mm or less is a short cervix, as SMFM’s Consult Series defines it, and every published definition applies only in the absence of cervical dilation. The threshold is an exact number and is applied exactly here, including the fact that 25 mm itself is short on that wording and 25.1 mm is not. Two preconditions sit in front of it. The measurement must be transvaginal: the Consult Series’ first recommendation is that any length used to guide treatment is measured that way, and transabdominal assessment at the anatomy scan is a screen rather than a measurement the thresholds were derived on. And the measurement has to be in the window the thresholds came from — the anatomy scan at 18 to 22 weeks, serial measurement from 16 to 24, treatment thresholds applying before 24 weeks.
What the number is worth depends far more on the history than on the millimetres. The positive predictive value of a short cervix for preterm birth before 34 to 35 weeks reaches 81.0% with a prior spontaneous preterm birth and is 26.3 to 39.1% in the general population: two to three times the predictive value from identical measurements. In nulliparas the numbers are starker still. At 25 mm, measured at 16 to 22 weeks, sensitivity for spontaneous preterm birth before 37 weeks is 8.0%, specificity 97.8%, and positive predictive value 16.2%. Tighten to 20 mm and sensitivity falls to 4.1% while specificity rises to 98.8% — which is what any tighter threshold does, and the page asserts the direction rather than asserting the figures alone. A short cervix is a highly specific and very insensitive test, and in an unselected population most preterm births happen to women whose cervix was never short.
Where the length is extreme the picture changes. Heath and colleagues found a length under 15 mm in 1.7% of an unselected population at 22 to 24 weeks, and that 1.7% accounted for 86% of preterm births before 28 weeks and 58% before 32. Where no length is measurable the Consult Series reports a 75% risk of delivery before 32 weeks with a median of three weeks from diagnosis — and says explicitly that not all such patients deliver before 32 weeks. Three things this page will not do. It will not apply the singleton thresholds to twins, because the Consult Series says in terms that there is no consensus on the measurement at which to diagnose a short cervix in twins. It will not hold a centile: cervical-length centiles for gestational age are a published table this calculator cannot interpolate, and the single useful figure — a 10th centile of about 26 mm at 18 to 24 weeks — is printed instead of a fitted curve. And it renders no management: the trial event rates quoted in the bands are what the studies found, not instructions. The companion measurement in labour is the Bishop score, which scores the cervix for a different purpose entirely, and in a twin pregnancy the growth discordance is the other question. This page computes a number, or applies published criteria, and names the body that publishes them. It renders no clinical decision: nothing here says whether this fetus is compromised, whether a tracing should be acted on, whether a scan should be repeated, whether labour should be induced or when a pregnancy should be delivered. An obstetric measurement is read alongside the whole clinical picture – history, gestational age, serial trend, the other biometry – and never instead of it. It supports a clinician’s judgement rather than replacing it. Thresholds here are the ones their named sources print. Reference standards differ between guidelines, a finding can cross a threshold purely by changing standard, and the cut-off your own unit works to takes precedence over anything on this page.
Frequently asked questions
Is 25 mm short, or does it have to be below 25?
SMFM’s Consult Series defines a midtrimester short cervix as 25 mm or less, so exactly 25.0 mm is short and 25.1 mm is not. This page implements that wording. Every definition also applies only in the absence of cervical dilation; a dilated cervix is a different finding that no length threshold describes.
Does a transabdominal measurement count?
Not for the published thresholds. SMFM’s first recommendation is that any cervical length used to guide treatment must be measured transvaginally. Transabdominal assessment is acceptable at the 18-to-22-week anatomy scan as a screen, but the thresholds were not derived on it, and this page refuses them rather than applying them to the wrong route.
Why does the obstetric history change the answer?
Because it changes the predictive value far more than the length does. The positive predictive value of a short cervix for preterm birth before 34 to 35 weeks reaches 81.0% with a prior spontaneous preterm birth and is 26.3 to 39.1% in the general population. The Consult Series read for this page is also scoped to patients without a prior spontaneous preterm birth, so its thresholds do not speak to that group and this page does not pretend they do.
What are the thresholds in twins?
There are none agreed. SMFM’s Consult Series states that in twin pregnancies there is no consensus on the measurement at which to diagnose a short cervix, and recommends against routine progesterone, pessary or cerclage outside clinical trials. The singleton numbers were not derived in twins and are not transferable.
Why is there no centile for gestational age?
Because the centiles are a published table and this calculator cannot interpolate one; a fitted curve would misclassify at the extremes, which is where the decision sits. The single figure worth having is that the 10th centile is about 26 mm at 18 to 24 weeks, which is why a length a little above 25 mm is not comfortably normal. The discrete thresholds, 25 mm, 20 mm, 15 mm and 10 mm, are exact published numbers and are applied exactly.
Related calculators
References
- Society for Maternal-Fetal Medicine. SMFM Consult Series #70: Management of short cervix in individuals without a history of spontaneous preterm birth. Am J Obstet Gynecol. 2024.
- Mountain Area Health Education Center Maternal-Fetal Medicine. Preterm Birth Prevention protocol, based on ACOG Practice Bulletin No. 234 and the April 2023 Practice Advisory. Asheville, NC.
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/