TABLE 2.

Characteristics of included studies on economic evaluation of prognostic prediction methods and multivariable models of caries or periodontitis

Reference	Sample	Prognostic prediction method and model		Economic evaluation			Results by authors
	Number (n) Age (year) Setting	Method/model Follow‐up time	Predictive performance	Study design	Type of evaluation: stated by authors as defined by Husereau et al. (2013)	Costs presented as
Prognostic prediction method for caries
Jokela and Pienihäkkinen (2003)	Group I Risk‐based prevention n = 299 Group II Routine prevention n = 226 2 years One group in each of two municipal health centers (1989–1993)	Group I: risk‐based prevention based on dental assistants' screening of mutans streptococci (MS) in proximal plaque (Dentocult SM® Strip Mutans) and incipient or other carious lesions low risk: MS‐free and caries‐free intermediate risk: MS‐positive with no caries high risk: signs of caries Group II: routine prevention based on dentists' decisions. Follow‐up time 3 years	Performance a to predict presence of cavitated lesions or fillings: Sensitivity: low‐risk versus intermediate + high‐risk 0.72 low‐risk + intermediate versus high‐risk 0.32 Specificity: low‐risk versus intermediate + high‐risk 0.77 low‐risk + intermediate versus high‐risk 0.98	Empirical study	Cost analysis Cost analysis	Mean running costs in Euros (€)	Cost per child for 3‐year time for examination, prevention and treatment: Group I 54€ Group II 69€ (Student's t‐test P = 0.004) Risk‐based prevention can be effective in reducing both costs and dental caries in preschool children when screening and preventive measures are delegated to dental assistants
Zavras et al. (2000)	n = 1180 1 or 2 years Community‐based private pediatric dental practice (1988–1995)	Group I: microbiological screening of salivary MS recorded as colony‐forming units (CFU) counts Group II: no screening Follow‐up time 1 or 2 years NR	Predictive performance based on CFU levels (low, moderate, high, too numerous to count) to predict caries: Sensitivity at 1 year 0.37–0.66 at 2 years 0.34–0.72 Specificity at 1 year 0.96–0.88 at 2 years 0.96–0.81	Model study	Cost analysis Cost analysis	Cost in US$ Costs based on fees for New England dental insurers	Cost for predictive method 20$ Total cost per child Group I 367.90$ Group II 396.70$ Cumulative dental treatment cost for a child aged 4 years lower if the child was screened Caries prevalence 15% (range 5%–51%) Cost savings increase significantly when caries prevalence increases
Multivariable model for prediction of caries
Holst and Braune (1994)	n = 102 2–4 years One test clinic in public dental health (1987–1991) versus. all public dental health clinics in one county (1991)	Risk assessment in test clinic based on factors given different weights: health status, medication, eating and drinking habits, oral hygiene, use of fluorides, parents knowledge of caries, parents interest in given information, visible caries Follow‐up times: from 2 to 4 years from 3 to 4 years Risk‐assessment by dental assistants and follow‐up examination by dentist	Predictive performance of risk assessment for manifest caries lesion: Sensitivity at 2–4 years 0.42 at 3–4 years 0.58 Specificity at 2–4 years 1.0 at 3–4 years 0.99	Empirical study	NR Cost‐effectiveness analysis	Mean time (min) spent per child up to 4 years at test clinic compared with mean time per child at clinics of the whole county based on county epidemiology	Mean value for time spent (min) at: test clinic 27 for dentist and 71 for dental assistant county clinics 60 for dentists and 90 for dental assistants Time spent was 50 min less in test clinic Caries prevalence 19% at test clinic and 23% at county clinics Test model for caries prevention is cost‐effective
Holst et al. (1997) b	n = 99 2–4 years n = 102 3–4 years One test clinic in public dental health (1990–1994) versus all public dental health clinics in one county (1994)	Risk assessment based on any single factor: illness for 1 week more than four times a year, saliva inhibiting drug, six daily intakes of food/drinks, anything else but water at night, oral hygiene less than once a day, no fluorides, visible plaque, visible caries Follow‐up times: from 2 to 4 years from 3 to 4 years Risk assessment by a dental assistant and follow‐up examination by dentist	Predictive accuracy of risk assessment for manifest caries lesions: Sensitivity at age 2–4 years 1.0 at age 3–4 years 0.86 Specificity at age 2–4 years 0.70 at age 3–4 years 0.66	Empirical study	NR Cost‐effectiveness analysis	Mean time minutes (min) spent per child up to 4 years at test clinic compared with mean time per child at county clinics based on county epidemiology	Mean value ‐time spent (min) at: test clinic 14 for dentist and 152 for dental assistant county clinics 42 for dentist and 102 for dental assistant Time spent for dentist was 28 min less in test clinic Caries prevalence 7% at test clinic and 24% at county clinics Test model for caries prevention is cost‐effective
Multivariable model for prediction of periodontitis
Higashi et al. (2002)	Hypothetical cohort with patients 35 years with mild periodontitis representing eight sub cohorts based on: treatment/no treatment smoker/non‐smoker Interleukin‐1 (IL‐1) genotype positive/negative Setting: periodontist specialist clinic	IL‐1 test (positive or negative) Follow‐up time: 30 years Periodontist	Predictive accuracy used for modeling assumption to identify patients with high risk for progression to severe periodontal disease: PPV 0.97 (range 0.94–1) NPV 0.97 (range 0.94–1)	Model study based on decision‐analysis and Markov modeling	Cost‐effectiveness analysis Cost‐utility analysis	Cost in US$ per Quality‐Adjusted Life‐Year (QALY)	Calculations of cost for genetic test 218$ Use of test compared with no‐test resulted in additional cost of 147,114$ per 1000 patients over a 30‐year time frame Reduction of number of cases with severe periodontitis 6.1 (absolute decrease 0.61%) QALYs increased by 4.5 using test Genetic test compared to no‐test ICER 32,633$ per QALY gained
Martin et al. (2014)	Group I patients receiving periodontal treatment n = 776 mean age 46 years (range 19–84) (1971–2003) Group II patients receiving routine dental care n = 523 males mean age 47.3 years (range 28–71) (1968–1988) Setting: private dental clinics	Chronic periodontitis (CP) risk score based on following factors: patient age, periodontal disease severity (deepest pocket, bleeding on probing, greatest radiographic bone loss), smoking history, diabetic status, periodontal treatment history, furcation involvements, vertical bone lesions, subgingival calculus or restorations Risk on a scale of 1 (very low risk) to 5 (very high risk) for alveolar bone loss and tooth loss Follow‐up: 13 years NR	Prediction of tooth loss c : Score 2 versus 3, 4, 5: Sensitivity 0.92 Specificity 0.33 PPV 0.59 NPV 0.80 LR+ 1.4 LR− 0.25 Score 2, 3 versus 4, 5: Sensitivity 0.60 Specificity 0.78 PPV 0.71 NPV 0.64 LR+ 2.7 LR− 0.51 Score 2, 3, 4 versus 5: Sensitivity 0.32 Specificity 0.93 PPV 0.83 NPV 0.56 LR+ 4.8 LR− 0.72	Model study	Cost–benefit analysis Cost‐effectiveness	Cost in US$ of periodontal treatment to preserve one tooth related to risk score and severity of CP	For high or moderate risk combined with any severity of CP, cost of periodontal treatment divided by number of teeth preserved ranged from 1405 to 4895$ Periodontal treatment is justified on basis of tooth preservation when risk is moderate or high regardless of CP severity For low risk with mild CP, cost of periodontal treatment is higher than fixed replacement

Abbreviations: ICER, incremental cost‐effectiveness ratio; LR, likelihood ratio; NPV, negative predictive value; NR, not reported; PPV, positive predictive value.

^aIn detail in Pienihäkkinen and Jokela (2002).

^bSimilar model as in study above. Different factors and extended sample.

^cAccording to our calculations based on Fig. 4 in Page et al. (2003).