Dental Literature Watch · Systematic review · Endodontics
Endodontically treated cracked teeth: 82% survive to 5 years, roughly 55-66% to 10, and deep probing along the crack triples the extraction hazard
TL;DR
- Pooled 4-to-5-year survival of endodontically treated cracked teeth was 82.2% (95% CI 67.7-91.1%; 7 cohorts; I² 89.5%).
- Beyond 5 years the curve keeps falling: two independent cohorts reported 10-year survival of 54% and 66%, and the Singapore cohort 55% at 15 years.
- Deep probing along the crack is the strongest quantified risk: in the longest cohort, adjusted extraction hazards were 2.48 for 4-5 mm and 3.19 for 6 mm or more.
- Other adverse markers: radicular extension of the crack, periapical disease, multiple cracks, older age, and no definitive cuspal coverage.
- Fifteen studies were included; nine contributed to quantitative synthesis. All were at serious risk of bias; certainty was very low.
- The authors' framing: prognosis should be time-dependent and individualised, not a single survival number.
The study
| Element | Detail |
|---|---|
| Design | Systematic review and meta-analysis of longitudinal cohorts; PubMed/Europe PMC, Embase and Web of Science searched to 3 October 2026 |
| Population | Endodontically treated permanent cracked teeth with at least 1 year of follow-up; primary outcome tooth survival |
| Synthesis | 4-to-5-year proportions pooled on the logit scale (random effects, REML, modified Knapp-Hartung); longer horizons and prognostic factors synthesised narratively; risk of bias by ROBINS-E-informed domains |
| Result: 4-5 years | 82.2% survival (95% CI 67.7-91.1%), 7 cohorts, I² 89.5% |
| Result: 10-15 years | 10-year survival 54% and 66% in two cohorts; 55% at 15 years in the Singapore cohort |
| Result: probing | Adjusted extraction hazard 2.48 for 4-5 mm and 3.19 for ≥ 6 mm crack-associated probing (longest cohort) |
| Certainty | All studies at serious risk of bias; very low certainty |
💎 Insight — what this means chairside
Probe the crack before you quote a number
A periodontal probe run along the crack line is the cheapest prognostic test in the review.
4-5 mm roughly two-and-a-half-times the extraction hazard; 6 mm or more roughly triples it.
A periodontal probe run along the crack line is the cheapest prognostic test in the review.
4-5 mm roughly two-and-a-half-times the extraction hazard; 6 mm or more roughly triples it.
Give consent in two time frames
"About four in five are still there at five years" is honest for a well-selected tooth.
"About half to two-thirds at ten" is the second sentence, and it is the one patients remember when the tooth fails.
"About four in five are still there at five years" is honest for a well-selected tooth.
"About half to two-thirds at ten" is the second sentence, and it is the one patients remember when the tooth fails.
Cuspal coverage is on the risk list, not the optional list
Lack of definitive cuspal coverage was an adverse marker across cohorts.
If the plan is root canal treatment, the plan includes the crown or onlay.
Lack of definitive cuspal coverage was an adverse marker across cohorts.
If the plan is root canal treatment, the plan includes the crown or onlay.
⚠️ The catch
Why this sits at 3 of 5
It sharpens one decision, whether to treat or extract a cracked tooth, and it does so with very-low-certainty data.
Nothing here is new in direction; the value is in putting numbers and a time axis on factors clinicians already weigh.
It sharpens one decision, whether to treat or extract a cracked tooth, and it does so with very-low-certainty data.
Nothing here is new in direction; the value is in putting numbers and a time axis on factors clinicians already weigh.
Heterogeneity of 89.5% and a wide interval
The 5-year interval runs from 67.7% to 91.1%, which is the difference between a good bet and a coin toss.
Settings, case selection and crack definitions varied, so your own practice may sit anywhere in that range.
The 5-year interval runs from 67.7% to 91.1%, which is the difference between a good bet and a coin toss.
Settings, case selection and crack definitions varied, so your own practice may sit anywhere in that range.
The 10-year figures come from two cohorts
54% and 66% are single-study numbers, not a pooled estimate.
The hazard ratios come from the longest cohort alone, so they describe one population's selection and treatment pattern.
54% and 66% are single-study numbers, not a pooled estimate.
The hazard ratios come from the longest cohort alone, so they describe one population's selection and treatment pattern.
Abstract only
Which cohorts, which teeth, what proportion were crowned, and how the crack was characterised are not in the abstract.
Which cohorts, which teeth, what proportion were crowned, and how the crack was characterised are not in the abstract.
⚡ Bottom line
- Cracked molar, shallow probing, no periapical lesion, cuspal coverage planned → around 82% at 5 years; say so, and say the 10-year figure is lower.
- Probing 6 mm or more along the crack → extraction hazard roughly tripled; put the implant or bridge conversation on the table now.
- Radicular extension, periapical disease, multiple cracks, older patient → each pulls the estimate down; stack them and extraction may be the kinder path.
- Patient asking at year 5 whether they are "out of the woods" → no; survival keeps declining, so keep cuspal coverage intact and keep probing.
🧪 Questions
Tap an answer — instant grade, deciding line, and the trap.
A 58-year-old has a cracked mandibular first molar with symptomatic irreversible pulpitis, no periapical lesion, and a 7 mm probing depth confined to the crack line. Which prognostic statement best reflects this review?
Deep crack-associated probing of at least 6 mm carried an adjusted extraction hazard of 3.19 in the longest cohort, so this tooth is a worse-than-average bet. Matching an uncracked tooth ignores that cracked teeth as a group already fall to roughly 55-66% by 10 years. Quoting 82% is the pooled average, which applies to the average case, not one with the strongest adverse marker present. Probing only with a periapical lesion is wrong because probing and periapical disease were independent risk indicators. Trap: the pooled 82%, which is true on average and misleading here.
A cracked premolar was root-filled and crowned five years ago and is symptom-free with 2 mm probing. The patient asks whether they are now out of the woods. Which answer does this review support?
Survival declines after 5 years, with 10-year figures of 54% and 66% and 55% at 15 years, so the review explicitly calls prognosis time-dependent. A plateau is the opposite of what the long cohorts show. Periapical status is one risk indicator, but its absence does not make survival stop falling. Tooth type was not among the listed risk indicators; age, radicular extension, multiple cracks, probing and cuspal coverage were. Trap: the plateau, which is how most patients and many clinicians intuit a five-year success.