Question · asked by 1 review
Multivariable caries prediction model vs Alternative prediction model(s) within the same study, or no formal comparator (models compared descriptively; no between-model statistical test): Predictive performance for root caries increment (negative likelihood ratio, LR-)
- Population
- Older adults (aged 52-95; mean 52-73) at risk of root caries increment
- Intervention
- Multivariable caries prediction model
- Comparison
- Alternative prediction model(s) within the same study, or no formal comparator (models compared descriptively; no between-model statistical test)
- Outcome
- Predictive performance for root caries increment (negative likelihood ratio, LR-)
The single root-caries validation study (Hayes 2017) showed LR- of 0.34-0.41, above the useful rule-out threshold; no pooled estimate; ungraded.
What each review found
Include earlier review versionsNo studies found 0 · 0%GRADE not reported 2 · 100%Very low 0 · 0%Low 0 · 0%Moderate 0 · 0%High 0 · 0%
| Finding | Effect | Studies | GRADE certainty |
|---|---|---|---|
Multivariable prediction model of caries increment (>=3 predictors), e.g. logistic-regression or algorithm-based caries-risk models vs Alternative prediction model(s) within the same study, or no formal comparator (models compared descriptively; no between-model statistical test) Predictive performance for root caries increment (negative likelihood ratio, LR-) · Older adults (aged 52-95; mean 52-73) at risk of root caries increment · Follow-up >=1 year (included studies 1-5 years), Oral health care; older adults; no geographical restriction, Model validation studies The single root-caries validation study (Hayes 2017) showed LR- of 0.34-0.41, above the useful rule-out threshold; no pooled estimate; ungraded. Multivariable prediction models of caries increment: a systematic review and critical appraisal. (2023) All outcomes | No meta-analysis. Hayes 2017 [33] (5 Cariogram root-caries models): per-model LR- (Table 2) ranged 0.34-0.41; none reached the useful rule-out threshold of <=0.20. | 1 study | GRADE not reported |
Multivariable prediction model of caries increment (>=3 predictors), e.g. logistic-regression or algorithm-based caries-risk models vs Alternative prediction model(s) within the same study, or no formal comparator (models compared descriptively; no between-model statistical test) Predictive performance for root caries increment (negative likelihood ratio, LR-) · Older adults (aged 52-95; mean 52-73) at risk of root caries increment · Follow-up >=1 year (included studies 1-5 years), Oral health care; older adults; no geographical restriction, Model development studies Developmental root-caries models showed modest negative likelihood ratios; only Powell 1991 reached the useful rule-out threshold; no pooled estimate; ungraded. Multivariable prediction models of caries increment: a systematic review and critical appraisal. (2023) All outcomes | No meta-analysis. Per-model LR- (Table 2): Ritter 0.36-0.48 (5 models), Powell 0.20 (0.04-0.90), Sanchez-Garcia 0.86 (0.80-0.93). Powell reached the useful rule-out threshold (LR- 0.20). | 2 studies | GRADE not reported |
Findings are shown side by side, never pooled across reviews.
The reviews
| Review | Searched up to | Included studies | AMSTAR 2 |
|---|---|---|---|
| Multivariable prediction models of caries increment: a systematic review and critical appraisal. 2023 All outcomes | 23 April 2021 | 22 | Not applicable |
Studies behind the findings
The primary studies the full-text reviews tie to this outcome. Reviews with headline outcomes only do not list them here.
- Evaluation of the Cariogram for root caries prediction. · 2017 · included by 1 of these review
- A Predictive Model for Root Caries Incidence. · 2016 · included by 1 of these review
- A prediction model for root caries in an elderly population. · 2011 · included by 1 of these review
- Exploration of prediction models for caries risk assessment of the geriatric population. · 1991 · included by 1 of these review
How reviews are matched to this question
Reviews word the same question differently. Four decision models (TypeSafe's Jev, Liquid AI's d1-ultra, SubQ Snap and Baseten's Elephant) judged each wording of the population, intervention, comparator and outcome against the others; two wordings count as the same when at least three of them agree and none clearly disagrees, and findings count as this question only when all four elements match. Timing, setting and subgroup never split a question: they are shown on each finding.