Abstract
Objective:
Quantify the survival of posterior composite restorations (PCR) placed during the study period in permanent teeth in United States (US) general dental community practices and factors predictive of that survival.
Methods:
A retrospective cohort study was conducted utilizing de-identified electronic dental record (EDR) data of patients who received a PCR in 99 general dentistry practices in the National Dental Practice-Based Research Network (Network). The final analyzed data set included 700,885 PCRs from 200,988 patients. Descriptive statistics and Kaplan Meier (product limit) estimator were performed to estimate the survival rate (defined as the PCR not receiving any subsequent treatment) after the first PCR was observed in the EDR during the study time. The Cox proportional hazards model was done to account for patient- and tooth-specific covariates.
Results:
The overall median survival time was 13.3 years. The annual failure rates were 4.5%–5.8% for years 1–5; 5.3%–5.7%, 4.9%–5.5%, and 3.3%–5.2% for years 6–10, 11–15, and 16–20, respectively. The failure descriptions recorded for <7% failures were mostly caries (54%) and broken or fractured tooth/restorations (23%). The following variables significantly predicted PCR survival: number of surfaces that comprised the PCR; having at least one interproximal surface; tooth type; type of prior treatment received on the tooth; Network region; patient age and sex. Based on the magnitude of the multivariable estimates, no single factor predominated.
Conclusions:
This study of Network practices geographically distributed across the US observed PCR survival rates and predictive factors comparable to studies done in academic settings and outside the US.
Keywords: Posterior composite restorations, retrospective studies, dental informatics, data science, electronic health record, electronic dental record
Introduction:
Resin-based composites (hereafter referred to as composites) have emerged during the last three decades as the standard direct restoration material [1,2]. A significant benefit of composites is the minimal removal of tooth structure due to their superior adhesive bonding to the underlying tooth structure. Thus, composites conserve coronal tooth structure weakened by caries or fractures. In addition, tooth-colored composites are desirable due to their superior esthetic properties, especially for direct restorations. Moreover, extensive evidence indicates that composite can be used as a restorative material in posterior teeth, resulting in long-term survival [1,3–11]. However, most of these studies originated from Netherlands and Brazil and no large-scale study evaluating posterior composite restorations (PCR) exists from the United States (US).
Early research on the survival of anterior and posterior composite restorations involved studying the durability and choice of appropriate restorative materials [12–14], bonding agents [15], and techniques, including the use of isolation such as rubber dam [16]. However, studies during the last 10–15 years have demonstrated that other factors also influence the long-term survival of composites, such as tooth type and location, patient characteristics (e.g., socioeconomic, demographic, and behavioral factors), and characteristics of the dentist who placed the initial restoration or assessed it during follow-up [17]. Material properties have a lesser influence on PCR longevity [17].
Systematic reviews on PCR survival, including clinical trials and retrospective studies with 5–33 years of follow-up, have reported annual failure rates of 0.08% to 4.9% [2,11,17,18]. In a recent systematic review, the majority (51.5%) of studies did not report restoration survival rates. Still, when reported, they ranged between 23% and 97.7% for follow-up periods from 5 – 30 years, with higher survival rates in studies with follow-up times from 5–10 years [11]. Studies included in these systematic reviews reported that caries and fractures were the main reasons for failure. The patients’ caries risk factors (e.g., oral hygiene, socioeconomic status, age, and sex) were also significant factors associated with PCR failure. In addition, operator-specific factors affect restoration survival, such as decisions to repair or replace the PCR. These results have demonstrated that characteristics related to restorations, restored teeth, patients, and operators have a more significant influence on restoration survival than restorative material characteristics.
During the last decade, numerous clinical studies using a broad range of observation periods have reported the survival of PCRs and factors that influence their clinical failure. About three-fourths of these studies were retrospective studies using electronic dental record (EDR) data in an academic clinic [11]. Most of these studies reported on treatment done outside the United States (US) [3,4,7–9,19–23]. Retrospective studies using EDR data have led to a better understanding of critical clinical events, such as caries, fractures, and loss of anatomical form that occur in restored teeth after many years. These results highlight the salience of performing such studies at the community practice level involving a representative population of everyday clinical practice.
To our knowledge, except for several prospective studies examining the benefits of repair versus replacement of restorations [24–27] and one study [28] evaluating restorations in older adults, no research has investigated the survival time of PCRs in US community practices. Therefore, this study investigated the survival time of PCRs performed in National Dental Practice-Based Research Network (Network) [29] general dentistry practices using their EDR data to determine factors that significantly predicted PCR survival time.
Methods:
We recruited 99 Network practices that consisted of small group and solo general dentistry practices who had used Dentrix® (Henry Schein One, American Fork, Utah) or EagleSoft® (Patterson Dental, St Paul, Minnesota) for at least five years to maintain electronic clinical documentation of at least existing conditions and treatment performed [30,31]. In addition, these practices performed at least one PCR on a permanent tooth in at least 100 patients, had follow-up data for at least two years and continued performing these procedures between January 1, 2000, and October 31, 2015. As detailed elsewhere [30], we received de-identified data of patients who had PCR treatment through October 2015 from 99 general dental practices. This data set included 203,212 patients with 834,802 PCR on permanent teeth identified using procedure codes (D2385, D2386, D2387, D2388, D2391, D2392, D2393, D2394). The PCR observed in the patient’s EDR during the study time is considered the initial PCR restoration and, therefore, the restoration followed in the study. These codes were also used to record glass ionomer restorations; consequently, the data set may include glass ionomer restorations for these posterior teeth. Although evidence in the literature is limited, reports indicate that composite is the most commonly used restorative material in the US [32]. In addition, the low flexural strength of glass ionomer material limits its use to restore posterior teeth surfaces [33,34]. The final data set to perform survival analysis required tooth number, surface(s), and dates of treatment to be present for a given PCR to be included in the dataset. Due to the study’s retrospective nature, we could not investigate the clinical acceptability of restorations and distinguish between repaired and replaced restorations.
Data analysis:
Descriptive statistics were performed to report patient demographics, the number of patient records included from the six Network regions, and the number of PCRs by number of surfaces. PCR failure (i.e., lack of survival) was defined as the PCR receiving any type of subsequent failure treatment during the follow-up period. The failure types included PCR subsequent to the initial PCR restoration during study time, crown, endodontic treatment, pulp capping, fixed prosthodontics, root extraction, surgical extraction, debridement hemisection, root amputation, apicoectomy, removable dentures, implant bodies, and fillings, inlays or onlays occurring on the same tooth and surface. In addition, conditions such as fractured or missing tooth were also considered failures. The number of surfaces was computed from the CDT codes and any restorations with ≥3 surfaces were combined as one group. A full list of ADA codes and conditions can be found in the appendix.
Kaplan Meier (product limit) estimator was performed to estimate the survival rate after the PCR performed on a specific tooth and specific surface(s). This method considered the right censored data that occurred if the patient did not have a subsequent failure on the tooth and surface. Cox proportional hazards survival analyses were used to test the associations of the patient’s age and sex, Network region, dental insurance, number of PCR surfaces, inclusion of at least one interproximal surface, tooth type, and other prior treatments done on the tooth (none, restoration, endodontic) with time to PCR failure. Restorations without procedures at a follow-up visit indicating PCR failure were censored at the time of the last visit for the patient. Analyses accounted for correlation among multiple PCRs within a patient using a robust sandwich estimate of the covariance matrix. Analyses were performed for each predictor individually and then in a multivariable model. A two-sided 5% significance level was used for all tests. All analyses were performed using R version 4.1.3 (R Foundation for Statistical Computing, Vienna, Austria) and SAS version 9.4 (SAS Institute, Inc, Cary, NC, USA). The methods described comply with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cohort studies.
Results:
The initial data set included 834,602 PCR observations in 203,212 patients, of which 707,617 PCRs were placed in a tooth by the respective practices. The final data set included 700,885 PCRs in 202,988 patients after removing 6,732 PCRs (unknown surface 2,335; unknown patient age 2,812; patient age prior to the eruption age of the tooth 1,560; inability to determine the treatment failure date 25). Table 1 shows the summary of the patient characteristics. Among 700,885 PCRs, 115,329 (16%) failed. Less than 7% (N=8,449) of these failures had a possible description of why the PCR failed (e.g. ‘decay’, taken from the description variable in the patient condition table as shown in the Appendix). The failure descriptions (N=8,449) included caries (54%), chipped/broken/fractured tooth (20%), poor/open margin (5%), cracked tooth (4%), missing/extracted tooth (3%), fractured/missing restoration (3%), and other reasons (11%) such as sensitivity/hypersensitivity, tooth wear, decalcification, irreversible pulpitis, endodontic treatment, overhanging margins, food impaction, recession and staining. Eighty-two combinations of failure types were tied specifically to the date of the PCR failure: 63% had only a PCR subsequent to the initial PCR restoration, 16% were crown only, 5% were root canal therapy only, and 5% were surgical extraction only; all other individual or combinations of failure types were identified in less than 5% of the patients.
Table 1.
Patient characteristics summary
| Characteristic | Level | N | (%) |
|---|---|---|---|
| Sex | Female | 114,515 | 56 |
| Male | 88,473 | 44 | |
| Total | 202,988 | 100 | |
| Region | Midwest | 25,639 | 13 |
| Northeast | 23,595 | 12 | |
| South Atlantic | 32,828 | 16 | |
| South Central | 45,261 | 22 | |
| Southwest | 45,920 | 23 | |
| Western | 29,745 | 15 | |
| Total | 202,988 | 100 | |
| Dental insurance | No | 44,846 | 22 |
| Yes | 76,917 | 38 | |
| unknown | 81,225 | 40 | |
| Total | 202,988 | 100 |
Characteristics of the patient, tooth, restoration, and practice summarized at the level of the restoration are shown in Table 2 for all PCRs combined and then separately for PCRs with/without failure. Eight percent of the PCRs involved both mesial and distal surfaces of the tooth, 35% involved either mesial or distal but not both (15% mesial only, 20% distal only), and 57% involved no interproximal surfaces. Follow-up times are also summarized in Table 2; 75% of the restorations were followed for 5 years or less, and maximum follow-up time was 31 years. Overall PCR survival using a Kaplan-Meier plot is shown in Figure 1. The annual failure rates (AFR) based on the Kaplan-Meier curves were relatively stable through 20 years. The overall yearly rates were 4.5%, 5.3%, 5.7%, 5.8%, and 5.8% for years 1–5; ranged from 5.3%–5.7%, 4.9%–5.5%, and 3.3%–5.2% for years 6–10, 11–15, and 16–20, respectively. The annual failure rates at years 1–5, 10, 15, and 20 are included in Table 4.
Table 2. Patient, tooth, restoration, and practice characteristics summary and follow-up time - time to failure or last follow-up at the restoration level.
Percentages within the 'All' column sum to 100% within the characteristic and represent the overall patient characteristic distributions. Percentages for follow-up time within the Not Failed and Failed columns also sum to 100%. Percentages sum to 100% across the Not Failed and Failed columns within each row for predictors of failure to represent the percentage of restorations that failed (or did not fail) within each level of a characteristic.
| Characteristics | Level | All N=700,885 (%) | Not Failed N=585,556 (%) | Failed N=115,329 (%) |
|---|---|---|---|---|
|
| ||||
| Age (years) | <18 | 136,817 (20) | 114,371 (84) | 22,446 (16) |
| 18 to 44 | 358,166 (51) | 310,651 (87) | 47,515 (13) | |
| 45 to 64 | 163,006 (23) | 127,861 (78) | 35,145 (22) | |
| 65 or older | 42,896 (6) | 32,673 (76) | 10,223 (24) | |
|
| ||||
| Sex | F | 395,572 (56) | 330,845 (84) | 64,727 (16) |
| M | 305.313 (44) | 254,711 (83) | 50,602 (17) | |
|
| ||||
| Dental insurance | No | 145,749 (21) | 123,491 (85) | 22,258 (15) |
| Yes | 286,780 (41) | 236,926 (83) | 49,854 (17) | |
| Unknown | 268,356 (38) | 225,139 (84) | 43,217 (16) | |
|
| ||||
| Tooth type | Premolar | 238,906 (34) | 201,471 (84) | 37,435 (16) |
| Molar | 461,979 (66) | 384,085 (83) | 77,894 (17) | |
|
| ||||
| Prior treatment | None | 548,261 (78) | 461,528 (84) | 86,733 (16) |
| Restoration | 133,875 (19) | 110,340 (82) | 23,535 (18) | |
| Endodontic | 18,749 (3) | 13,688 (73) | 5,061 (27) | |
|
| ||||
| Number of Surfaces | 1 | 314,910 (45) | 266,725 (85) | 48,185 (15) |
| 2 | 112,826 (16) | 228,977 (84) | 44,172 (16) | |
| >=3 | 273,149 (39) | 89,854 (80) | 22,972 (20) | |
|
| ||||
| At least 1 interproximal surface | No | 398,490 (57) | 338,819 (85) | 59,671 (15) |
| Yes | 302,395 (43) | 246,737 (82) | 55,658 (18) | |
|
| ||||
| Network region | Midwest | 79,744 (11) | 64,200 (85) | 15,544 (15) |
| Northeast | 90,924 (13) | 68,738 (84) | 22,186 (16) | |
| South Atlantic | 100,840 (14) | 85,355 (86) | 15,485 (14) | |
| South Central | 159,375 (23) | 134,531 (86) | 24,844 (14) | |
| South west | 159,082 (23) | 137,146 (84) | 21,936 (16) | |
| Western | 110,920 (16) | 95,586 (87) | 15,334 (13) | |
|
| ||||
| Follow-up time | No follow up | 77,500 (11) | 77,500 (13) | |
|
| ||||
| Up to 1 year | 158,905 (23) | 135,498 (23) | 23,407 (20) | |
|
| ||||
| >1 & ≤5 years | 301,329 (43) | 238,624 (41) | 62,705 (54) | |
|
| ||||
| >5 & ≤10 years | 123,058 (18) | 98,924 (17) | 24,134 (21) | |
|
| ||||
| >10 & ≤15 years | 35,308 (5) | 30,611 (5) | 4,697 (4) | |
|
| ||||
| >15 & ≤20 years | 4,647 (1) | 4,285 (1) | 362 (<1) | |
|
| ||||
| >20 years | 138 (<1) | 114 (<1) | 24 (<1) | |
Figure 1.

Overall Kaplan-Meier survival plot for time to PCR failure
Table 4.
Median survival times of posterior composite restorations (PCRs) from the multivariable survival analysis* and annual failure rates from Kaplan-Meier curves.
| Predictor | Median Survival (years), 95% CI | 1 yr | 2 yr | 3 yr | 4 yr | 5 yr | 10 yr | 15 yr |
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| Restoration-level | ||||||||
|
| ||||||||
| Surfaces in the PCR | ||||||||
| 1 | 14.5 (14.2–14.7) | 3.6% | 4.7% | 4.9% | 5.1% | 5.0% | 4.8% | 4.2% |
| 2 | 12.9 (12.8–13.1) | 4.6% | 5.6% | 6.0% | 6.1% | 6.3% | 5.8% | 5.2% |
| >=3 | 11.4 (11.2–11.6) | 6.9% | 6.7% | 7.2% | 7.3% | 7.3% | 7.1% | 7.3% |
|
| ||||||||
| PCR at least 1 interproximal surface | ||||||||
| No | 14.9 (14.7–15.2) | 3.6% | 4.8% | 5.1% | 5.2% | 5.2% | 4.8% | 4.4% |
| Yes | 11.6 (11.4–11.7) | 5.7% | 6.2% | 6.5% | 6.6% | 6.8% | 6.5% | 5.9% |
|
| ||||||||
| Tooth-level | ||||||||
|
| ||||||||
| Tooth type in which PCR placed | ||||||||
| Premolar | 15.7 (15.5–16.0) | 4.2% | 4.9% | 5.1% | 5.2% | 5.3% | 4.8% | 4.6% |
| Molar | 12.2 (12.1–12.3) | 4.6% | 5.6% | 6.0% | 6.1% | 6.1% | 5.9% | 5.1% |
|
| ||||||||
| Type of prior treatment, if any, in the tooth that received the PCR | ||||||||
| None | 13.7 (13.6–13.9) | 4.0% | 5.1% | 5.5% | 5.7% | 5.7% | 5.3% | 4.9% |
| Restoration | 12.8 (12.6–13.0) | 5.2% | 5.4% | 5.9% | 5.9% | 6.0% | 6.3% | 5.1% |
| Endodontic | 6.9 (6.7–7.1) | 13% | 12% | 9.3 % | 9.2% | 8.6% | 6.5% | 7.0% |
|
| ||||||||
| Patient-level | ||||||||
|
| ||||||||
| Age of patient with PCR | ||||||||
| Pediatric | 11.2 (11.0–11.4) | 3.0% | 6.5% | 7.3% | 7.3% | 7.4% | 5.7% | 4.6% |
| 18 to 44 | 15.3 (15.0–15.5) | 4.0% | 4.2% | 4.6% | 4.8% | 5.0% | 5.2% | 4.8% |
| 45 to 64 | 12.6 (12.4–12.8) | 5.8% | 5.7% | 5.8% | 6.0% | 5.8% | 5.7% | 5.5% |
| 65 or Older | 9.6 (9.4–9.8) | 7.6% | 8.3% | 8.0% | 7.9% | 7.4% | 5.9% | 2.0% |
|
| ||||||||
| Dental insurance status of patient with the PCR | ||||||||
| No | 13.2 (13.0–13.4) | 5.4% | 5.6% | 5.7% | 5.7% | 5.3% | 5.5% | 4.5% |
| Yes | 13.3 (13.1–13.4) | 4.3% | 5.1% | 5.4% | 5.5% | 5.7% | 5.4% | 4.7% |
| Unknown | 13.3 (13.1–13.5) | 4.2% | 5.5% | 5.9% | 6.1% | 6.2% | 5.5% | 5.6% |
|
| ||||||||
| Sex of patient with the PCR | ||||||||
| F | 13.3 (13.2-13.5) | 4.6% | 5.2% | 5.5% | 5.7% | 5.7% | 5.2% | 4.9% |
| M | 13.2 (13.0–13.4) | 4.3% | 5.5% | 5.8% | 5.9% | 5.9% | 5.8% | 4.9% |
|
| ||||||||
| Practice-level * | ||||||||
|
| ||||||||
| Region of practice that placed the PCR | ||||||||
| Midwest | 13.1 (12.9–13.4) | 4.1% | 5.0% | 5.9% | 6.1% | 6.2% | 6.1% | 6.8% |
| Northeast | 11.1 (10.9–11.2) | 5.0% | 7.2% | 7.4% | 7.5% | 7.1% | 6.1% | 5.0% |
| South Atlantic | 13.0 (12.8–13.2) | 5.2% | 5.1% | 5.2% | 5.2% | 5.3% | 5.6% | 5.5% |
| South Central | 13.4 (13.2–13.6) | 4.2% | 5.3% | 5.8% | 5.8% | 5.6% | 4.9% | 4.2% |
| Southwest | 14.2 (14.0–14.5) | 4.4% | 4.7% | 4.7% | 4.8% | 5.3% | 4.9% | 3.8% |
| Western | 14.2 (13.9–14.5) | 4.2% | 5.0% | 5.3% | 5.6% | 5.6% | 5.4% | 7.8% |
Mean survival value is not reported because the censoring used in a survival analysis can create bias in estimates of means.
From the results of the multivariable survival analysis, at the restoration level (Table 3 and Figure 2), time to PCR failure decreased from 1 to 2 to ≥3 surfaces (p<0.0001). Inclusion of at least one interproximal surface reduced the time to failure (p<0.0001). Molars were at higher risk of earlier failure than premolars (p<0.0001). Prior treatment on the tooth was associated with a higher risk of failure (p<0.0001), with endodontic treatment having the shortest time to failure. In comparison, previously restored teeth also had a slightly higher risk of failure than previously untreated teeth. At the patient level, time to PCR failure was shortest for patients 65 years and older (p<0.0001), followed by pediatric patients, patients 45–64, and patients 18–44 years. Insurance (p=0.89) and sex (p=0.19) were not significantly associated with time to PCR failure. Time to failure also differed between regions, with the shortest time to failure in the Northeast (p<0.0001), followed by South Atlantic and Midwest, and South Central, with the longest survival in the Western and Southwest regions. Median survival times, with 95% confidence intervals, for the levels of each factor are shown in Table 4.
Table 3.
Associations of restoration-level, tooth-level, patient-level, and practice-level predictors with time to posterior composite restoration (PCR) failure
| Bivariate Survival Analyses | Multivariable Survival Analysis | |||
|---|---|---|---|---|
|
| ||||
| Predictor | p-value | Hazard Ratio (95% CI) | p-value | Hazard Ratio (95% CI) |
|
| ||||
| Restoration-level | ||||
|
| ||||
| Number of surfaces in the PCR | <.0001 | <.0001 | ||
| >=3 vs 2 | <.0001 | 1.25 (1.23–1.27) | <.0001 | 1.14 (1.12–1.16) |
| >=3 vs 1 | <.0001 | 1.54 (1.51–1.57) | <.0001 | 1.27 (1.23–1.30) |
| 2 vs 1 | <.0001 | 1.23 (1.21–1.25) | <.0001 | 1.11 (1.09–1.14) |
|
| ||||
| PCR has at least 1 interproximal surface | <.0001 | <.0001 | ||
| Yes vs No | <.0001 | 1.35 (1.33–1.37) | <.0001 | 1.29 (1.26–1.32) |
|
| ||||
| Tooth-level | ||||
|
| ||||
| Tooth type in which PCR placed | <.0001 | <.0001 | ||
| Molar vs Premolar | <.0001 | 1.15 (1.13–1.17) | <.0001 | 1.28 (1.26–1.30) |
|
| ||||
| Type of prior treatment, if any, in the tooth that received the PCR | <.0001 | |||
| Endodontic vs Restoration | <.0001 | 1.84 (1.78–1.91) | <.0001 | 1.88 (1.82–1.95) |
| Endodontic vs None | <.0001 | 2.06 (2.00–2.13) | <.0001 | 2.02 (1.95–2.08) |
| Restoration vs None | <.0001 | 1.12 (1.10–1.14) | <.0001 | 1.07 (1.05–1.09) |
|
| ||||
| Patient-level | ||||
|
| ||||
| Age of patient with PCR | <.0001 | <.0001 | ||
| 65 or Older vs 45 to 64 | <.0001 | 1.32 (1.28–1.36) | <.0001 | 1.31 (1.28–1.35) |
| 65 or Older vs 18 to 44 | <.0001 | 1.65 (1.60–1.70) | <.0001 | 1.58 (1.54–1.63) |
| 65 or Older vs Pédiatrie | <.0001 | 1.31 (1.27–1.36) | <.0001 | 1.16 (1.12–1.20) |
| 45 to 64 vs 18 to 44 | <.0001 | 1.25 (1.22–1.27) | <.0001 | 1.20 (1.18–1.23) |
| Pediatrie vs 45 to 64 | 0.5807 | 1.01 (0.98–1.03) | <.0001 | 1.13 (1.10–1.16) |
| Pediatrie vs 18 to 44 | <.0001 | 1.26 (1.23–1.29) | <.0001 | 1.36 (1.33–1.39) |
|
| ||||
| Dental insurance of the patient with PCR | <.0001 | 0.8866 | ||
| No vs Yes | < 0001 | 1.06 (1.03–1.08) | 1.00 (0.98–1.03) | |
| No vs Unknown | 0.9876 | 1.00 (0.98–1.02) | 1.01 (0.98–1.03) | |
| Yes vs Unknown | <0001 | 1.06 (1.04–1.08) | 1.00 (0.98–1.02) | |
|
| ||||
| Sex of patient with the PCR | 0.0038 | 0.1872 | ||
| M vs F | 0.0038 | 1.02 (1.01–1.04) | 1.01 (0.99–1.03) | |
|
| ||||
| Practice-level | ||||
| Region of practice that paced the PCR | <.0001 | <.0001 | ||
| Northeast vs Midwest | <.0001 | 1.21 (1.17–1.25) | <.0001 | 1.19 (1.16–1.23) |
| South Atlantic vs Midwest | 0.1755 | 1.02 (0.99–1.05) | 0.4649 | 1.01 (0.98–1.04) |
| Midwest vs South central | 0.0031 | 1.04 (1.01–1.07) | 0.1655 | 1.02 (0.99–1.05) |
| Midwest vs Southwest | <.0001 | 1.13 (1.10–1.16) | <.0001 | 1.08 (1.05–1.11) |
| Midwest vs Western | 0.0004 | 1.06 (1.02–1.09) | <.0001 | 1.08 (1.05–1.11) |
| Northeast vs South Atlantic | <.0001 | 1.23 (1.20–1.27) | <.0001 | 1.18 (1.14–1.21) |
| Northeast vs South Central | <.0001 | 1.26 (1.23–1.29) | <.0001 | 1.21 (1.18–1.25) |
| Northeast vs Southwest | <.0001 | 1.37 (1.33–1.40) | <.0001 | 1.29 (1.25–1.33) |
| Northeast vs Western | <.0001 | 1.28 (1.24–1.31) | <.0001 | 1.28 (1.25–1.32) |
| South Atlantic vs South Central | 0.1299 | 1.02 (0.99–1.05) | 0.0263 | 1.03 (1.00–1.06) |
| South Atlantic vs Southwest | <.0001 | 1.11 (1.08–1.14) | <.0001 | 1.09 (1.06–1.13) |
| South Atlantic vs Western | 0.0269 | 1.03 (1.00–1.07) | <.0001 | 1.09 (1.06–1.12) |
| South Central vs Southwest | <.0001 | 1.08 (1.06–1.11) | <.0001 | 1.06 (1.03–1.09) |
| South Central vs Western | 0.3679 | 1.01 (0.99–1.04) | 0.0001 | 1.06 (1.03–1.09) |
| Western vs Southwest | <.0001 | 1.07 (1.04–1.10) | 0.8008 | 1.00 (0.98–1.03) |
Figure 2.


Kaplan-Meier survival plots for associations of each characteristic individually with time to PCR failure
Discussion:
This study demonstrates an overall median survival time of 13.3 years for the 700,885 PCRs (202,988 patients) performed in 99 US community practices across the six Network regions. Overall, 16% of the restorations failed during follow-up. Twenty percent of these failures occurred within 1 year and another 54% failed between 1 and 5 years after placement. As reported in previous studies, the number of surfaces that comprised the restoration significantly predicted restoration longevity; restorations comprising 3 or more surfaces had the shortest median (11.4 years) survival time (Table 4). Similarly, restorations that included an interproximal surface had a median survival time of 11.6 years. History of prior treatment was strongly associated with a shorter survival time; teeth with an endodontic treatment history failed earliest (see Table 4). The importance of ongoing caries risk assessment and management even after restoration placement is suggested by the high percentage of failures due to caries (53%) and broken/fractured tooth/restorations (23%) among the failed restorations for which reasons were available. These findings are comparable to other studies performed outside the United States [1,2,5,7,11,14,22,23] that indicated restoration-, tooth-, patient- and practice-level factors influence longevity of restorations. Nevertheless, this study contributes evidence regarding the survival time of PCRs placed in general dentistry Network practices in the US. In addition, this study utilized EDR data that contained 700,885 PCRs in patients regardless of their socioeconomic and insurance statuses and performed in practices distributed across the six Network regions. Previous practice-based restoration longevity studies that employed large datasets were based on claims data [9,35,36] that allow only study of the insured population. Therefore, this study could be considered to have a higher generalizability because the study data originated from dental care visits of all patients irrespective of their insurance status [7,30]. Some studies [7,37] have shown that restorations last longer in females. However, in this study, patients’ sex was not significantly associated with PCR longevity, similar to other studies, including a recent systematic review [11,23]. However, patient age was significantly associated with PCR longevity; restorations placed in patients 65 years, and older had the shortest median survival time of 9.6 years, followed by the pediatric population with permanent teeth (Table 4). Also, the presence or absence of dental insurance was not significantly associated with the longevity of PCRs. However, this finding is inconclusive because the dental insurance of 40% of this study cohort could not be determined.
The overall AFR of 4.5% to 5.8% for years 1–5 and 5.3% to 5.7% for years 6–10 years, are comparable to other PBR studies that analyzed data from multiple dental practices in Netherlands [7,8]. Another multiple practice-based study with 11 practices showed a lower AFR of 2.5% for posterior composite restorations at 10 years [38]. However, the AFR for teeth with prior endodontic treatment was less at 5 years compared to these studies but higher in years 1 and 2. Moreover, restorations placed in the Northeast Network practices had a higher AFR (7.1–7.6%) compared to other regions in years 2–5 (Table 4). This difference could be due to the larger number of practices that participated in this study and the geographical distribution of these practices across the six Network regions nationally in the US (Tables 3 and 4). The diverse characteristics of the patient populations (e.g., dental insurance status or socioeconomic status) that this large national sample included may have also influenced the study results. In addition, not distinguishing between repair and replacement of restorations in the failures may have also reduced the survival rates in this study [11]. The regional differences in survival that we observed across the Network regions could reflect regional differences among dentists’ evaluation and treatment decisions regarding when to replace versus repair a restoration, which then influences PCR longevity. Therefore, future research should investigate regional differences in the diagnosis and decision-making process during dental care and define thresholds for failure and reintervention [11] and replacing or repairing a restoration. For instance, a PCR may be left alone if it is functional but not aesthetically acceptable unless the patient complains about its appearance [11].
Limitations:
We could not differentiate between composite and glass ionomer restorations because practices use the same procedure code for restorations placed using these materials. Second, the first restoration identified in the patient’s EDR during the study may not be the first restoration placed on the tooth [28]. Third, we could not evaluate other factors that might influence restoration longevity, such as systemic diseases, number and type of medications taken, oral conditions (e.g., dry mouth, opposition dentition, bruxism, removable denture design use, and oral hygiene habits) and dental materials (composite type, etchant characteristics and technique, isolation methods, light curing duration)[11,17,28]. We also could not distinguish between repaired and replaced restorations. Finally, only about 25% of restorations had a follow-up time of more than five years because patients tend to change practices when they move or fail to return to the same practice for other reasons.
Failure descriptions for only <7 percent of the failed restorations signify dental clinicians’ need to record treatment reasons as a diagnostic code or in the clinical notes. Such information would enable clinicians to improve their assessment and treatment planning based on their work [7,22]. The rapid advances in natural language processing methods allow retrieval of pertinent information from clinical notes, provided clinicians ensure recording them [39–42].
Conclusions:
This study identified the major predictors of PCRs longevity when performed in US community practices. Tooth-level factors (number of restored surfaces, presence of an interproximal surface, history of prior treatment, especially endodontic treatment and tooth type) were significant factors associated with longevity of PCRs. However, at the patient level, while the time to PCR failure was shortest among patients 65 years and above, followed by children, time to failure was not significantly associated with patients’ dental insurance or sex. The results could be informative for patients and general dentists during the diagnosis and treatment planning phases of dental care.
Supplementary Material
Acknowledgments
This work was supported by NIH grants U19-DE-28717 and U19-DE-22516. Opinions and assertions contained herein are those of the authors and are not to be construed as necessarily representing the views of the respective organizations or the National Institutes of Health. This study was determined not to be human subjects research by the Indiana University Institutional Review Board (IRB). Approvals or exemptions were also obtained separately from the six network regions’ IRBs. The manuscript co-authors reported having no conflicts of interest. An Internet site devoted to details about the nation’s network is located at http://NationalDentalPBRN.org.
We gratefully acknowledge the practitioner participants who shared their data to conduct this study, without which this study would not have been possible. We also gratefully acknowledge the assistance of Mr. Neil Butler and Chris Wadsworth from Henry Schein and Mr. Ted Fruchtl and Ms. Jessica Knaus from Patterson Dental with extracting data from the network practices and the support of these two companies’ leadership for the study. We thank the Regenstrief Institute’s staff for their assistance with the project.
We are grateful to the network’s regional node coordinators, who function as the “face” of the network in recruiting, training, and interacting with the network’s practitioners (Midwest Region: Tracy Shea, RDH, BSDH; Western Region: Stephanie Hodge, MA; Northeast Region: Christine O’Brien, RDH; South Atlantic Region: Hanna Knopf, BA, and Deborah McEdward, RDH, BS, CCRP; South Central Region: Shermetria Massengale, MPH, CHES, and Ellen Sowell, BA; Southwest Region: Stephanie Reyes, BA, Meredith Buchberg, MPH, and Colleen Dolan, MPH), and the network’s program manager (Andrea Mathews, BS, RDH), and program coordinator (Terri Jones).
Footnotes
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
CRediT author statement
Thankam Thyvalikakath: conceptualization, funding acquisition, supervision, methodology, data acquisition, analysis and interpretation, drafted and critically revised the manuscript for important intellectual content.
Zasim Azhar Siddiqui: data acquisition, analysis and interpretation, drafted initial manuscript and gave final approval of the manuscript.
George Eckert: methodology, data analysis and interpretation, critically revised and gave final approval of the manuscript.
Michelle LaPradd: data curation, analysis and interpretation, final approval of the manuscript.
William D. Duncan: methodology, data acquisition, data curation, final approval of the manuscript.
Valeria V Gordan: supervision, methodology, data acquisition, interpretation and gave final approval of the manuscript.
D Brad Rindal: supervision, methodology, data acquisition, interpretation and gave final approval of the manuscript.
Mark Jurkovich: methodology, data acquisition, analysis, interpretation and gave final approval of the manuscript.
Gregg Gilbert: conceptualization, funding acquisition, supervision, methodology, data acquisition, analysis, and interpretation, critically revised the manuscript for important intellectual content and gave final approval of the manuscript.
National Dental PBRN Collaborative Group: methodology, data acquisition, interpretation and gave final approval of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Thankam Thyvalikakath, Office of Dental Informatics & Digital Health, Indiana University School of Dentistry, IUPUI; Dental Informatics, Regenstrief Institute, Inc., OH 144A, 415 Lansing Street, Indianapolis, IN 46202, Indianapolis, IN 46202, USA.
Zasim Azhar Siddiqui, Current affiliation: Doctoral student, West Virginia University School of Pharmacy, Morgantown, West Virginia, USA; Affiliation during study: Postdoctoral Fellow, Dental Informatics, Department of Cariology, Operative Dentistry & Dental Public Health, Indiana University School of Dentistry, IUPUI, Indianapolis, IN 46202, USA.
George Eckert, Department of Biostatistics and Health Data Science, Indiana University School of Medicine, 340 W 10th St, Indianapolis, IN 46202, USA.
Sr. Michelle LaPradd, Current affiliation: Syneos Health, 1030 Sync St, Morrisville, NC 27560, USA; Affiliation during study: Staff Biostatistician, Department of Biostatistics and Health Data Science, Indiana University School of Medicine, 340 W 10th St, Indianapolis, IN 46202, USA.
William D. Duncan, Current affiliation: Department of Community Dentistry, University of Florida, College of Dentistry, Gainesville, FL, USA; Affiliation during study: Associate Director, Biomedical Data Science and Shared Resource, Roswell Park Cancer Center, Buffalo, NY, USA.
Valeria V Gordan, Dental Practice-Based Research, Interim Associate Dean for Research, University of Florida, College of Dentistry, Gainesville, FL, USA.
D. Brad Rindal, HealthPartners Institute, HealthPartners Dental Group, 8170 33rd Avenue South | P.O. Box 1524, MS 23301A Minneapolis MN 55440, United States.
Mark Jurkovich, HealthPartners Institute, Minneapolis MN, 8170 33rd Ave S, Bloomington, MN 55440, United States.
Gregg H. Gilbert, National Dental Practice-Based Research Network, Department of Clinical and Community Sciences, School of Dentistry, SDB Room 109, University of Alabama at Birmingham, Birmingham, AL, United States.
National Dental PBRN Collaborative Group, 1720 University Blvd, Birmingham, AL 35294, University of Alabama at Birmingham, Birmingham, AL, United States.
References
- [1].Da Rosa Rodolpho PA, Rodolfo B, Collares K, Correa MB, Demarco FF, Opdam NJM, Cenci MS, Moraes RR, Clinical performance of posterior resin composite restorations after up to 33 years, Dent Mater. 38 (2022) 680–688. 10.1016/j.dental.2022.02.009. [DOI] [PubMed] [Google Scholar]
- [2].Maciel CM, Baroudi K, Barroso LDS, Costa LDC, Souto TCV, Pino Vitti R, Longevity of Resin Composite and Amalgam Posterior Restorations: A Systematic Review, Eur J Prosthodont Restor Dent. (2022). 10.1922/EJPRD_2371Maciel09. [DOI] [PubMed] [Google Scholar]
- [3].Opdam NJM, Bronkhorst EM, Roeters JM, Loomans BAC, A retrospective clinical study on longevity of posterior composite and amalgam restorations., Dental Materials : Official Publication of the Academy of Dental Materials. 23 (2007) 2–8. 10.1016/j.dental.2005.11.036. [DOI] [PubMed] [Google Scholar]
- [4].Opdam NJM, Bronkhorst EM, Loomans BAC, Huysmans MCDNJM, 12-year survival of composite vs. amalgam restorations, Journal of Dental Research. 89 (2010) 1063–1067. 10.1177/0022034510376071. [DOI] [PubMed] [Google Scholar]
- [5].Opdam NJM, van de Sande FH, Bronkhorst E, Cenci MS, Bottenberg P, Pallesen U, Gaengler P, Lindberg A, Huysmans MCDNJM, van Dijken JW, Longevity of Posterior Composite Restorations: A Systematic Review and Meta-analysis, J Dent Res. 93 (2014) 943–949. 10.1177/0022034514544217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Pallesen U, van Dijken JWV, A randomized controlled 30 years follow up of three conventional resin composites in Class II restorations, Dent Mater. 31 (2015) 1232–1244. 10.1016/j.dental.2015.08.146. [DOI] [PubMed] [Google Scholar]
- [7].Laske M, Opdam NJM, Bronkhorst EM, Braspenning JCC, Huysmans MCDNJM, Ten-Year Survival of Class II Restorations Placed by General Practitioners, JDR Clinical & Translational Research. 1 (2016) 292–299. 10.1177/2380084416663192. [DOI] [PubMed] [Google Scholar]
- [8].Laske M, Opdam NJM, Bronkhorst EM, Braspenning JCC, Huysmans MCDNJM, Longevity of direct restorations in Dutch dental practices. Descriptive study out of a practice based research network, Journal of Dentistry. 46 (2016) 12–17. 10.1016/j.jdent.2016.01.002. [DOI] [PubMed] [Google Scholar]
- [9].Burke FJT, Lucarotti PSK, The ultimate guide to restoration longevity in England and Wales. Part 4: resin composite restorations: time to next intervention and to extraction of the restored tooth, Br Dent J. 224 (2018) 945–956. 10.1038/sj.bdj.2018.443. [DOI] [PubMed] [Google Scholar]
- [10].Montag R, Dietz W, Nietzsche S, Lang T, Weich K, Sigusch BW, Gaengler P, Clinical and Micromorphologic 29-year Results of Posterior Composite Restorations, J Dent Res. 97 (2018) 1431–1437. 10.1177/0022034518788798. [DOI] [PubMed] [Google Scholar]
- [11].Demarco FF, Cenci MS, Montagner AF, de Lima VP, Correa MB, Moraes RR, Opdam NJM, Longevity of composite restorations is definitely not only about materials, Dental Materials. 39 (2023) 1–12. 10.1016/j.dental.2022.11.009. [DOI] [PubMed] [Google Scholar]
- [12].Knobloch LA, Kerby RE, Seghi R, Berlin JS, Clelland N, Fracture toughness of packable and conventional composite materials, J Prosthet Dent. 88 (2002) 307–313. 10.1067/mpr.2002.128069. [DOI] [PubMed] [Google Scholar]
- [13].Yap AUJ, Soh MS, Post-gel polymerization contraction of “low shrinkage” composite restoratives, Oper Dent. 29 (2004) 182–187. [PubMed] [Google Scholar]
- [14].Mr. Abreu N, Sousa FB, Dantas RV, Kb. Leite P, Batista AU, Montenegro RV, Longevity of bulk fill and ormocer composites in permanent posterior teeth: Systematic review and meta-analysis, Am J Dent. 35 (2022) 89–96. [PubMed] [Google Scholar]
- [15].Costa T, Rezende M, Sakamoto A, Bittencourt B, Dalzochio P, Loguercio AD, Reis A, Influence of Adhesive Type and Placement Technique on Postoperative Sensitivity in Posterior Composite Restorations, Oper Dent. 42 (2017) 143–154. 10.2341/16-010-C. [DOI] [PubMed] [Google Scholar]
- [16].Soncini JA, Maserejian NN, Trachtenberg F, Tavares M, Hayes C, The longevity of amalgam versus compomer/composite restorations in posterior primary and permanent teeth: findings From the New England Children’s Amalgam Trial, J Am Dent Assoc. 138 (2007) 763–772. 10.14219/jada.archive.2007.0264. [DOI] [PubMed] [Google Scholar]
- [17].Demarco FF, Corrêa MB, Cenci MS, Moraes RR, Opdam NJM, Longevity of posterior composite restorations: Not only a matter of materials, Dental Materials. 28 (2012) 87–101. 10.1016/j.dental.2011.09.003. [DOI] [PubMed] [Google Scholar]
- [18].da Veiga AMA, Cunha AC, Ferreira DMTP, da Silva Fidalgo TK, Chianca TK, Reis KR, Maia LC, Longevity of direct and indirect resin composite restorations in permanent posterior teeth: A systematic review and meta-analysis, Journal of Dentistry. 54 (2016) 1–12. 10.1016/j.jdent.2016.08.003. [DOI] [PubMed] [Google Scholar]
- [19].Wierichs RJ, Kramer EJ, Meyer-Lueckel H, Risk Factors for Failure of Direct Restorations in General Dental Practices, J Dent Res. 99 (2020) 1039–1046. 10.1177/0022034520924390. [DOI] [PubMed] [Google Scholar]
- [20].Collares K, Opdam NJM, Laske M, Bronkhorst EM, Demarco FF, Correa MB, Huysmans MCDNJM, Longevity of Anterior Composite Restorations in a General Dental Practice-Based Network, Journal of Dental Research. 96 (2017) 1092–1099. 10.1177/0022034517717681. [DOI] [PubMed] [Google Scholar]
- [21].Palotie U, Eronen AK, Vehkalahti K, Vehkalahti MM, Longevity of 2- and 3-surface restorations in posterior teeth of 25- to 30-year-olds attending Public Dental Service-A 13-year observation., Journal of Dentistry. 62 (2017) 13–17. 10.1016/j.jdent.2017.05.012. [DOI] [PubMed] [Google Scholar]
- [22].Laske M, Opdam NJM, Bronkhorst EM, Braspenning JCC, Huysmans MCDNJM, Risk Factors for Dental Restoration Survival: A Practice-Based Study, J Dent Res. 98 (2019) 414–422. 10.1177/0022034519827566. [DOI] [PubMed] [Google Scholar]
- [23].van de Sande FH, Opdam NJ, Rodolpho PADR, Correa MB, Demarco FF, Cenci MS, Patient risk factors’ influence on survival of posterior composites, J Dent Res. 92 (2013) 78S–83S. 10.1177/0022034513484337. [DOI] [PubMed] [Google Scholar]
- [24].Gordan VV, Garvan CW, Richman JS, Fellows JL, Rindal DB, Qvist V, Heft MW, Williams OD, Gilbert GH, How dentists diagnose and treat defective restorations: Evidence from the dental Practice-Based Research Network, Operative Dentistry. 34 (2009) 664–673. 10.2341/08-131-C. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Gordan VV, Riley JL, Rindal DB, Qvist V, Fellows JL, Dilbone DA, Brotman SG, Gilbert GH, Repair or replacement of restorations: A prospective cohort study by dentists in The National Dental Practice-Based Research Network, The Journal of the American Dental Association. 146 (2015) 895–903. 10.1016/j.adaj.2015.05.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Gordan VV, Riley J, Geraldeli S, Williams OD, Spoto JC, Gilbert GH, The decision to repair or replace a defective restoration is affected by who placed the original restoration: Findings from the National Dental PBRN, Journal of Dentistry. 42 (2014) 1528–1534. 10.1016/j.jdent.2014.09.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Gordan VV, Garvan CW, Heft MW, Fellows JL, Qvist V, Rindal DB, Gilbert GH, DPBRN Collaborative Group, Restorative treatment thresholds for interproximal primary caries based on radiographic images: findings from the Dental Practice-Based Research Network., General Dentistry. 57 (2009) 654–63; quiz 664–6, 595, 680. [PMC free article] [PubMed] [Google Scholar]
- [28].Caplan DJ, Li Y, Wang W, Kang S, Marchini L, Cowen HJ, Yan J, Dental Restoration Longevity among Geriatric and Special Needs Patients, JDR Clinical & Translational Research. 4 (2019) 41–48. 10.1177/2380084418799083. [DOI] [PubMed] [Google Scholar]
- [29].Gilbert GH, Fellows JL, Allareddy V, Cochran DL, Cunha-Cruz J, Gordan VV, McBurnie MA, Meyerowitz C, Mungia R, Rindal DB, National Dental PBRN Collaborative Group, Structure, function, and productivity from the National Dental Practice-Based Research Network, J Clin Transl Sci. 6 (2022) e87. 10.1017/cts.2022.421. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Thyvalikakath TP, Duncan WD, Siddiqui Z, LaPradd M, Eckert G, Schleyer T, Rindal DB, Jurkovich M, Shea T, Gilbert GH, National Dental PBRN Collaborative Group, Leveraging Electronic Dental Record Data for Clinical Research in the National Dental PBRN Practices, Appl Clin Inform. 11 (2020) 305–314. 10.1055/s-0040-1709506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Thyvalikakath T, LaPradd M, Siddiqui Z, Duncan WD, Eckert G, Medam JK, Rindal DB, Jurkovich M, Gilbert GH, National Dental PBRN Collaborative Group, Root Canal Treatment Survival Analysis in National Dental PBRN Practices, J Dent Res. 101 (2022) 1328–1334. 10.1177/00220345221093936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Worthington HV, Khangura S, Seal K, Mierzwinski-Urban M, Veitz-Keenan A, Sahrmann P, Schmidlin PR, Davis D, Iheozor-Ejiofor Z, Rasines Alcaraz MG, Direct composite resin fillings versus amalgam fillings for permanent posterior teeth, Cochrane Database Syst Rev. 8 (2021) CD005620. 10.1002/14651858.CD005620.pub3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Ge KX, Quock R, Chu C-H, Yu OY, The preventive effect of glass ionomer restorations on new caries formation: A systematic review and meta-analysis, Journal of Dentistry. 125 (2022) 104272. 10.1016/j.jdent.2022.104272. [DOI] [PubMed] [Google Scholar]
- [34].Su N-Y, Wang Y-H, Chang Y-C, A registry-based study of tooth-colored restorative materials for decayed teeth in Taiwan, Journal of Dental Sciences. 18 (2023) 1235–1242. 10.1016/j.jds.2023.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].Lucarotti PSK, Burke FJT, The ultimate guide to restoration longevity in England and Wales. Part 1: methodology, Br Dent J. 224 (2018) 709–716. 10.1038/sj.bdj.2018.267. [DOI] [PubMed] [Google Scholar]
- [36].Raedel M, Hartmann A, Bohm S, Priess H-W, Samietz S, Konstantinidis I, Walter MH, Four-year outcomes of restored posterior tooth surfaces-a massive data analysis, Clin Oral Investig. 21 (2017) 2819–2825. 10.1007/s00784-017-2084-4. [DOI] [PubMed] [Google Scholar]
- [37].Rho Y-J, Namgung C, Jin B-H, Lim B-S, Cho B-H, Longevity of direct restorations in stress-bearing posterior cavities: a retrospective study, Oper Dent. 38 (2013) 572–582. 10.2341/12-432-C. [DOI] [PubMed] [Google Scholar]
- [38].Casagrande L, Laske M, Bronkhorst EM, Huysmans MCDNJM, Opdam NJM, Repair may increase survival of direct posterior restorations - A practice based study, J Dent. 64 (2017) 30–36. 10.1016/j.jdent.2017.06.002. [DOI] [PubMed] [Google Scholar]
- [39].Patel JS, Kumar K, Zai A, Shin D, Willis L, Thyvalikakath TP, Developing Automated Computer Algorithms to Track Periodontal Disease Change from Longitudinal Electronic Dental Records, Diagnostics (Basel). 13 (2023) 1028. 10.3390/diagnostics13061028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [40].Patel JS, Shin D, Willis L, Zai A, Kumar K, Thyvalikakath TP, Comparing gingivitis diagnoses by bleeding on probing (BOP) exclusively versus BOP combined with visual signs using large electronic dental records, Sci Rep. 13 (2023) 17065. 10.1038/s41598-023-44307-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [41].Büttner M, Leser U, Schneider L, Schwendicke F, Natural Language Processing: Chances and Challenges in Dentistry, J Dent. 141 (2023) 104796. 10.1016/j.jdent.2023.104796. [DOI] [PubMed] [Google Scholar]
- [42].VanSchaik JT, Jain P, Rajapuri A, Cheriyan B, Thyvalikakath TP, Chakraborty S, Using transfer learning-based causality extraction to mine latent factors for Sjögren’s syndrome from biomedical literature, Heliyon. 9 (2023) e19265. 10.1016/j.heliyon.2023.e19265. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
