JDR JDR Most Cited Articles
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (30)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chuang, S.K.
Right arrow Articles by Dodson, T.B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chuang, S.K.
Right arrow Articles by Dodson, T.B.
J Dent Res 81(8): 572-577, 2002
© 2002 International and American Associations for Dental Research

RESEARCH REPORT
Clinical

Risk Factors for Dental Implant Failure: A Strategy for the Analysis of Clustered Failure-time Observations

S.K. Chuang1,*, L.J. Wei2, C.W. Douglass3, and T.B. Dodson4

1 Department of Oral Health Policy and Epidemiology, Harvard School of Dental Medicine, 188 Longwood Avenue, Boston, MA 02115;
2 Department of Biostatistics, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115; and
3 Department of Oral Health Policy and Epidemiology, Harvard School of Dental Medicine, 188 Longwood Avenue, Boston, MA 02115;
4 Department of Oral and Maxillofacial Surgery, Harvard School of Dental Medicine and Massachusetts General Hospital, 55 Fruit Street, Warren 1201, Boston, MA 02114;

* corresponding author, PO Box 67376, Chestnut Hill Station, Chestnut Hill, MA 02467, schuang{at}hsph.harvard.edu


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS & METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
This study’s objective was to identify, in a statistically valid and efficient manner, the risk factors associated with dental implant failure. We hypothesize that factors exist which can be modified by clinicians to enhance outcome. A retrospective cohort study design was used. Cohort members had >= one implant placed. Risk factors were classified as demographic, health status, implant-, anatomic-, or prosthetic-specific, and reconstructive variables. The outcome variable was implant failure. The cohort was composed of 677 patients who had 2349 implants placed. Based on the adjusted multivariate model, factors associated with implant failure were tobacco use, implant length, staging, well size, and immediate implants (p <= 0.05). In the setting of correlated survival observations, we recommend adjusting for the correlation of the observations to provide statistically valid and efficient results. Three of the identified factors—tobacco use, immediate implants, and implant staging—potentially may be modified to enhance implant survival.

KEY WORDS: survival analysis • dental implants • risk factors • follow-up study • correlation and dependence • Cox regression analysis • clustered survival data • marginal approach


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS & METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The endosseus dental implant is a predictable technology to facilitate the prosthetic replacement of teeth. The focus of implant research is shifting from descriptions of clinical success to the identification of factors associated with failure (Esposito et al., 1999). A better understanding of the factors associated with implant failure provide data for the planning of future studies, facilitate clinical decision-making, and may enhance implant success.

To date, most studies evaluating risk factors for failure are flawed in terms of their statistical analyses. Many researchers assessed survival in a binary manner (yes or no) (Jemt et al., 1996; Lazzara et al., 1996; Rosenquist and Grenthe, 1996; Cooper et al., 1999; Chaffee et al., 2002) or applied statistical methods assuming that the implant observations were independent of each other (Wheeler, 1996; Buser et al., 1997; Brocard et al., 2000; Testori et al., 2001). Reporting survival as a binary outcome tends to overestimate survival, because long-term failures are diluted by the early success of recently placed implants (Eckert and Wollan, 1998).

As a practical clinical and research issue, it is common for patients to have more than one dental implant, thus violating the key assumption of independent observations (Mau, 1993; Haas et al., 1996). To address the issue of correlated, dependent observations, Haas et al. (1996), Lekholm et al. (1999), and Herrmann et al. (1999) recommended randomly selecting one implant per patient for analysis. While this solution works, inefficient estimation occurs, since not all observations are used at the same time during sampling.

Valid and efficient clustered survival statistical techniques applied to the identification of factors associated with implant failure have not been investigated extensively. Eckert and co-workers (Eckert and Wollan, 1998; Eckert et al., 2001) utilized the robust standard error method of Wei et al. (1989) for adjusting for possible dependence due to multiple implants per subject. It is our opinion that the Eckert studies (Eckert and Wollan, 1998; Eckert et al., 2001) applied the method incorrectly. It would be more appropriate to apply the clustered survival methodologies described by Lee et al. (1992), Lin (1994), and Spiekerman and Lin (1998). Spiekerman and Lin (1998), using advanced theoretical survival methodologies, provided more rigorous and complete proofs than the previous two papers mentioned.

This study’s specific aim was to identify risk factors associated with implant failure by applying a clustered failure-time multivariate model. We believe that this is the first report with integrated clinical applications of innovative theoretical clustered survival methodologies to identify risk factors associated with dental implant failure (Lee et al., 1992); Lin, 1994; Spiekerman and Lin, 1998). In addition, we hypothesized that some factors identified in this analysis may be modified by the clinician to enhance implant survival.


   MATERIALS & METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS & METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Study Design/Source Population
To address our specific aims, we conducted a retrospective cohort study. The study cohort had heterogeneous risk factors (Rothman and Greenland, 1998) and was derived from the population of patients who had >= one Bicon® implant inserted at the Implant Dentistry Centre at Faulkner Hospital (IDC-FH, Boston, MA) between May 20, 1992, and July 6, 2000. All patients who had implants placed at the IDC-FH were eligible for inclusion in the study. The IDC-FH is a teaching facility. This study was approved by the institutional human studies review committee (Protocol #: 1999-P-011145).

Criteria for study inclusion were as follows: (1) All surgical treatment was completed at the IDC-FH; and (2) all patients—regardless of medical health status, age, gender, race, or abilities—were included. Exclusion criteria included inadequate or unavailable patient charts.

Study Variables
Risk factors associated with implant failure were grouped into the following categories: demographic; health status; anatomic-, implant- and abutment-specific; anticipated prosthetic restoration; peri-operative chemotherapy; and reconstructive. Demographic variables included age and gender.

    Health-status variables
General health status was classified according to the American Society of Anesthesiology (ASA) system (Dental Implant Clinical Research Group, 1997). Patients were categorized as healthy (ASA 1), as having mild systemic disease (ASA 2), or as having moderate or severe systemic disease (ASA 3). In addition, we recorded whether the patient had a medical condition that may compromise wound healing, such as immunosuppression or diabetes, and current tobacco use status (American Society of Anesthesiologists, 1998).

    Anatomic variables
These included implant position (maxilla, mandible, anterior, posterior), bone quality (Types 1–4), and proximity of the implant relative to other teeth or implants. The relationship of the implant to other dento-alveolar structures was grouped into the following categories: no teeth (edentulous), one natural tooth, two natural teeth, one implant, two implants, one natural tooth, and one implant (Truhlar et al., 1997).

    Implant-specific variables
These included size (width 3 to 6 mm, length 6 to 14 mm), coating (uncoated, titanium-plasma-sprayed [TPS], hydroxyapatite [HA]), abutment size (in terms of diameter, length, and angulations), or use of a temporary implant. Prosthetic variables were grouped into removable (overdenture), fixed denture or bridge, or single-crown restorations.

    Peri-operative chemotherapy variables
These included the type, dose, and frequency of chemotherapeutic agents, i.e., antibiotics and chlorhexidine (Dent et al., 1997; Lambert et al., 1997).

    Reconstructive variables
These included types and materials used for augmentation and timing of implant placement relative to the augmentation procedure. The timing of fixture placement was categorized as immediate (reconstructive procedure and implant placed on same day) or delayed (reconstructive procedure and implant placement separated by time).

We recorded the dates of the following clinical events: implant placement, abutment placement, permanent restoration placement, and last visit or implant removal, where applicable. The major outcome variable of interest was implant failure. Failure was defined as the removal of the implant for any reason (Dental Implant Clinical Research Group, 1997). Total survival time would be the duration of time (months) from implant placement to implant removal or date of last follow-up for patients whose implants had not been removed.

Statistical Issues
We modeled implant failure time with the Cox proportional hazards model with no specific dependence structure among implants. The regression coefficients, ß, in the model were estimated by , which maximized the "partial likelihood" function L(ß) obtained by assuming that the observations were independent. Although observations may be correlated due to matching or clustering, was still consistent and asymptotically normal (Lee et al., 1992; Spiekerman and Lin, 1998). The standard variance-covariance matrix estimate, the inverse of -{partial}2logL()/{partial}ß2, however, may no longer be valid for inferences about ß. We have utilized a valid variance-covariance matrix estimate, which accounted for the dependence among related dental implant observations within the same subject (Lee et al., 1992; Lin, 1994). Spiekerman and Lin (1998) provided a vigorous survival methodology to fill in some important gaps in the existing proofs of the two papers mentioned previously. They have shown that the mathematical vector of regression parameters under the independence working assumption was consistent and asymptotically normal, with a variance-covariance matrix for which a consistent estimator was provided. It should be noted that our procedures were similar to those studied by Liang and Zeger (1986) for non-censored observations with ordinary likelihood for longitudinal data analysis.

Inference Procedures for Clustered Survival Marginal Models
Let Tik denote the failure time of the kth implant in the ith individual, k = 1,2,3,....,Ki; i = 1,2,3,...,n. We assumed that Ki was relatively small with respect to n, the total number of patients in the study. Also, we assumed that maxiKi was bounded (< {infty}). Let Zik be a p x 1 vector of bounded risk factors and Cik be the censoring variable. For Tik, one observed a bivariate vector (Xik, {delta}ik), where Xik = min(Tik, Cik) and {delta}ik = 1, if Tik = Xik was observed, 0 otherwise. The censoring vector C'i = (Ci1,...,CiKi)', i = 1,...,n are censoring vectors that were assumed to be independent of the failure time variables T'i = (Ti1,...,TiKi)' and Z'i = (Zi1,...,ZiKi). Let Zik (t) denote a p x 1 vector of risk factors for Tik at time t > 0. We supposed that conditional on Zik = zik, the marginal hazard function {lambda}ik (t) for Tik had the usual proportional hazards form:

, where {lambda}0 (t) was an arbitrary hazard function, and ß0 denoted the vector of the true regression coefficients.

We could obtain an estimator of ß0 based on the working assumption that the dental implants in each individual were independent of one another. Under this assumption, the logarithm methodological set-up of the "partial likelihood" loge(ß) function discussed by Lee et al. (1992, p. 239) for the dental implant application was as follows:


where Yjk(t) = Number of [Xjk >= t]. Note that the estimator is the one that maximizes loge(ß). Although the dental implant observations in each individual may be correlated, Lee et al. (1992) have shown mathematically that, under rather mild conditions, was still consistent for ß0. This was also shown by Spiekerman and Lin (1998) using more rigorous asymptotic theory for the estimation of the regression parameters. In our dental implant applications, for large n, the distribution of n1/2 ( - ß0) can be approximated by a normal distribution with mean 0 and variance-covariance matrix {Gamma}, which was also discussed by Lee et al. (1992). If the dental implants in each individual were indeed independent of each other, the matrix was the robust variance-covariance matrix estimate for proposed by Lin and Wei (1989).

In brief, the key approach formulated the marginal distributions of multivariate failure times with the Cox proportional hazards models while leaving the nature of dependence among related failure times completely unspecified. The estimating equations for the regression parameters developed by Lin (1994) and further explored vigorously by Spiekerman and Lin (1998) yielded consistent and asymptotically normal estimators. We constructed the robust variance-covariance estimators to account for the intra-class correlation among the dental implants in the same individual to produce valid and efficient statistical inferences.

Data Management and Analysis
We created a database using EpiInfo 2000 (Centers for Disease Control and Prevention, Atlanta, GA) with appropriate checks to identify errors. Descriptive statistics were computed for all study variables. Univariate analyses were used to identify risk factors associated with survival. Risk factors with p-values <= 0.15 based on univariate analyses and biologically relevant variables were entered into a multivariate marginal Cox proportional hazards regression model that adjusted for clustering failure-time observations. Advanced survival statistical computing methodologies used the S-plus (Version 3.4, Math Soft 1996) programming environment in the Unix operating system.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS & METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
During the study period, 701 patients were eligible for study enrollment. Charts for 24 patients were not available for review because the patients were either deceased, had moved, were inactive, or could not be found in the computer, or the chart could not be found. The final study cohort was composed of 677 (96.6%) patients who had 2349 implants placed. There was no evidence of systematic loss of charts.

The cohort was composed of patients with a mean age of 53.1 ± 13.8 yrs, and 50.4% were women. Most patients ({approx} 99%) were healthy or had mild systemic disease (ASA scores <= 2), and 10.3% reported tobacco use at the time the implant was placed. The mean duration of follow-up was 23.8 mos (range, 0.3 to 90.9 mos). The descriptive statistics for all of the study variables are summarized in Table 1Go. The overall one- and five-year Kaplan-Meier survival estimates with associated 95% confidence intervals adjusted for clustered observations were 95.4% (95% CI: 94.2%, 96.6%) and 91.2% (95% CI: 88.8%, 93.6%), respectively (Chuang et al., 2001). Clinical manifestations of implant failure were primarily inflammatory, e.g., mobility, pain, infection, or peri-implantitis.


View this table:
[in this window]
[in a new window]
  		Table 1. Descriptive Statistics (total n = 677 patients; total k = 2349 dental implants)
 
Table 2Go summarizes the univariate relationships between the study variables and implant failure. Variables associated with implant failure (p <= 0.15) were current tobacco use, history of tobacco use, anatomic location of the implant, implant length, coating, well size, prosthetic type, abutment diameter, position of the implant relative to other teeth or implants, immediate placement of implants, and implant stage.


View this table:
[in this window]
[in a new window]
  		Table 2. Univariate Analysis for Risk Factors Associated with Implant Failure (total n = 677 patients; total k = 2349 dental implants)a
 
Clustered parsimonious multivariate Cox marginal regression models were developed that accounted for the clustering effect of implants within the same subject (Lee et al., 1992; Lin, 1994). Variables included in the multivariate model were selected because of biologic importance, age, and sex, or were statistically associated with implant failure in univariate analyses (p <= 0.15) (see Table 3Go). In the parsimonious multivariate Cox model, tobacco use, implant length, well size, immediate implant, and implant staging remained statistically associated with implant failure (p <= 0.05) (see Table 4Go).


View this table:
[in this window]
[in a new window]
  		Table 3. Multivariate Marginal Cox Regression Model (adjusted) (total n = 677 patients; total k = 2349 dental implants)a
 

View this table:
[in this window]
[in a new window]
  		Table 4. Parsimonious Multivariate Marginal Cox Regression Model (adjusted) (total n = 677 patients; total k = 2349 dental implants)a
 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS & METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Most statistical analyses of dental implant survival data are flawed or inefficient. Many analyses incorrectly assume that multiple implants placed in a single patient can be treated as independent observations. Violating the independence assumption renders statistical inferences invalid. An alternative strategy is to select randomly one implant per patient for analysis. Such a strategy produces statistically valid results; however, it is inefficient because it wastes data.

The purpose of this study was to identify risk factors associated with implant failure in a statistically valid and efficient manner. Innovative multivariate Cox marginal regression models were developed and applied that accounted for the clustering effect of multiple implants within the same subject (Lee et al., 1992; Lin, 1994). In brief, current tobacco use (hazard ratio = 3.1), longer implant length (hazard ratio = 0.7), immediate implant placement after tooth or implant removal (hazard ratio = 1.8), staging of the implant (hazard ratio = 0.3), and wider well size (hazard ratio = 0.3) were significantly (p <= 0.05) associated with implant failure. Consistent with our proposed hypothesis, several of the above-listed variables, i.e., tobacco use, implant staging, and immediate implant placement, may potentially be manipulated by the clinician to enhance outcome. For example, for patients who smoke, the clinician can elect not to place the implant or to inform the patient of the consequences of smoking on implant survival.

Retrospective cohort studies rely on the completeness of data entered into the patient’s chart. Data may be missing because of misplaced, misfiled, discarded, or missing information in the chart. There is no reason to believe, however, that these records or entries in the progress notes were selectively missing because of the presence or absence of the key variables. Fortunately, chart entries were made by a small number of staff members at the implant center; however, the possibility of incomplete recording of information persists, but is thought not to be selectively incomplete. Retrospective cohort studies have less validity than randomized prospective clinical trials, due to issues of selection bias and confounding. Appropriate clustered multivariate Cox regression analyses are required for better control of confounding risk factors. Study design remains important. Despite the robust statistical analyses for failure time data, the conclusions that can be drawn from retrospective cohort studies may be limited.

In summary, this paper identified multiple risk factors associated with implant failure, including smoking status, implant length, immediate implant placement, implant staging, and well size. In the setting of correlated survival observations, we recommend adjusting for the correlation of the observations to provide statistically valid and efficient estimates of the parameters for risk factors under investigation.


   ACKNOWLEDGMENTS
 
This research is supported in part by Dentist Scientist Award NIH/NIDCR K16 DE000275 (SKC), NIH/NCI grant R01 CA56844 (LJW), and Mid-Career Investigator Award in Patient-Oriented Research, NIH/NIDCR K24 DE000448 (TBD). Dr. Dodson is also supported by the Oral and Maxillofacial Surgery Research Fund, Massachusetts General Hospital (TBD). The abstract of this manuscript was selected as an AADR Hatton competition finalist in the senior category held at San Diego, California, during the IADR/AADR meetings from March 5 to March 9, 2002, for the first author (SKC). This manuscript partially fulfilled the requirements for the doctorate degree (DMSc) at Harvard University for the first author (SKC). We also thank Ms. Valerie Vehemente for her assistance in data collection. The authors recognize the clinicians and staff of the Implant Dentistry Centre at the Faulkner Hospital, Boston, MA, for their cooperation in this study and their free and unfettered access to patient records. We also thank the reviewers for their suggestions for the revision of this manuscript.

Received November 5, 2001; Last revision May 31, 2002; Accepted June 5, 2002


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS & METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
American Society of Anesthesiologists (ASA) (1998). Physical status classes. In: Clinical anesthesia procedures of the Massachusetts General Hospital. 5th 3d. Philadelphia: Lippincott-Raven.

Brocard D, Barthet P, Baysse E, Duffort JF, Eller P, Justumus P, et al. (2000). A multicenter report on 1,022 consecutively placed ITI implants: a 7-year longitudinal study. Int J Oral Maxillofac Implants 15:691–700.[Medline]

Buser D, Mericske-Stern R, Bernard JP, Behneke A, Behneke N, Hirt HP, et al. (1997). Long-term evaluation of non-submerged ITI implants. Part 1: 8-year life table analysis of a prospective multi-center study with 2359 implants. Clin Oral Implants Res 8:161–172.[Medline]

Chaffee NR, Felton DA, Cooper LF, Palmqvist U, Smith R (2002). Prosthetic complications in an implant-retained mandibular overdenture population: initial analysis of a prospective study. J Prosthet Dent 87:40–44.[Medline]

Chuang SK, Tian L, Wei LJ, Dodson TB (2001). Kaplan-Meier analysis of dental implant survival: a strategy for estimating survival with clustered observations. J Dent Res 80:2016–2020.[Abstract/Free Full Text]

Cooper LF, Scurria MS, Lang LA, Guckes AD, Moriarty JD, Felton DA (1999). Treatment of edentulism using Astra Tech implants and ball abutments to retain mandibular overdentures. Int J Oral Maxillofac Implants 14:646–653.[Medline]

Dent CD, Olson JW, Farish SE, Bellome J, Casino AJ, Morris HF, et al. (1997). The influence of preoperative antibiotics on success of endosseous implants up to and including stage II surgery: a study of 2,641 implants. J Oral Maxillofac Surg55(Suppl 5):19–24.[Medline]

Dental Implant Clinical Research Group (1997). Introduction. J Oral Maxillofac Surg 55(Suppl 5):7–11.[Medline]

Eckert SE, Wollan PC (1998). Retrospective review of 1170 endosseous implants placed in partially edentulous jaws. J Prosthet Dent79:415–421.[Medline]

Eckert SE, Meraw SJ, Weaver AL, Lohse CM (2001). Early experience with Wide-Platform Mk II implants. Part I: Implant survival. Part II: Evaluation of risk factors involving implant survival. Int J Oral Maxillofac Implants 16:208–216.[Medline]

Esposito M, Hirsch J, Lekholm U, Thomsen P (1999). Differential diagnosis and treatment strategies for biologic complications and failing oral implants: a review of the literature. Int J Oral Maxillofac Implants 14:473–490.[Medline]

Haas R, Mensdorff-Pouilly N, Mailath G, Watzek G (1996). Survival of 1,920 IMZ implants followed for up to 100 months. Int J Oral Maxillofac Implants11:581–588.[Medline]

Herrmann I, Lekholm U, Holm S, Karlsson S (1999). Impact of implant interdependency when evaluating success rates: a statistical analysis of multicenter results. Int J Prosthodont 12:160–166.[Medline]

Jemt T, Chai J, Harnett J, Heath MR, Hutton JE, Johns RB, et al. (1996). A 5-year prospective multicenter follow-up report on overdentures supported by osseointegrated implants. Int J Oral Maxillofac Implants11:291–298.[Medline]

Lambert PM, Morris HF, Ochi S (1997). The influence of 0.12% chlorhexidine digluconate rinses on the incidence of infectious complications and implant success. J Oral Maxillofac Surg 55(Suppl 5):25–30.[Medline]

Lazzara R, Siddiqui AA, Binon P, Feldman SA, Weiner R, Phillips R, et al. (1996). Retrospective multicenter analysis of 3i endosseous dental implants placed over a five-year period. Clin Oral Implants Res 7:73–83.[Medline]

Lee EW, Wei LJ, Amato DA (1992). Cox-type regression analysis for large numbers of small groups of correlated failure time observations. In: Survival analysis: state of the art. Klein JP, Goel PK, editors. Dordrecht: Kluwer Academic, pp. 237-247.

Lekholm U, Gunne J, Henry P, Higuchi K, Linden U, Bergstrom C, et al. (1999). Survival of the Brånemark implant in partially edentulous jaws: a 10-year prospective multicenter study. Int J Oral Maxillofac Implants 14:639–645.[Medline]

Liang KY, Zeger SL (1986). Longitudinal data analysis using generalized linear models. Biometrika 73:13–22.[Abstract/Free Full Text]

Lin DY (1994). Cox regression analysis of multivariate failure time data: the marginal approach. Stat Med 13:2233–2247.[Medline]

Lin DY, Wei LJ (1989). The robust inference for the Cox proportional hazards model. J Am Stat Assoc 84:1074–1078.

Mau J (1993). On statistics of success and loss for dental implants. Int Dent J 43:254–261.[Medline]

Rosenquist B, Grenthe B (1996). Immediate placement of implants into extraction sockets: implant survival. Int J Oral Maxillofac Implants 11:205–209.[Medline]

Rothman KJ, Greenland S (1998). Modern epidemiology. 2nd ed. Philadelphia: Lippincott-Raven Publishers.

S-Plus (1996). StatSci, A division of MathSoft, Inc. Version 3.4, Seattle, WA.

Spiekerman CF, Lin DY (1998). Marginal regression models for multivariate failure time data. J Am Stat Assoc 93:1164–1175.

Testori T, Wiseman L, Woolfe S, Porter SS (2001). A prospective multicenter clinical study of the Osseotite implant: four-year interim report. Int J Oral Maxillofac Implants 16:193–200.[Medline]

Truhlar RS, Orenstein IH, Morris HF, Ochi S (1997). Distribution of bone quality in patients receiving endosseous dental implants. J Oral Maxillofac Surg 55(Suppl 5):38–45.[Medline]

Wei LJ, Lin DY, Weissfeld L (1989). Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc 84:1065–1073.

Wheeler SL (1996). Eight-year clinical retrospective study of titanium plasma-sprayed and hydroxyapatite-coated cylinder implants. Int J Oral Maxillofac Implants 11:340–350.[Medline]




This article has been cited by other articles:


Home page
 J. Dent. Res.Home page
S.-K. Chuang and T. Cai
Predicting Clustered Dental Implant Survival Using Frailty Methods.
J. Dent. Res., December 1, 2006; 85(12): 1147 - 1151.
[Abstract] [Full Text] [PDF]

Home page
 J. Dent. Res.Home page
M.C.M. Wong, K.F. Lam, and E.C.M. Lo
Bayesian Analysis of Clustered Interval-censored Data
J. Dent. Res., September 1, 2005; 84(9): 817 - 821.
[Abstract] [Full Text] [PDF]

Home page
 J Dent EducHome page
M. M. Walsh and J. A. Ellison
Treatment of Tobacco Use and Dependence: The Role of the Dental Professional
J Dent Educ., May 1, 2005; 69(5): 521 - 537.
[Abstract] [Full Text] [PDF]

Home page
 J. Dent. Res.Home page
S.K. Chuang, T. Cai, C.W. Douglass, L.J. Wei, and T.B. Dodson
Frailty Approach for the Analysis of Clustered Failure Time Observations in Dental Research
J. Dent. Res., January 1, 2005; 84(1): 54 - 58.
[Abstract] [Full Text] [PDF]

Home page
 J. Dent. Res.Home page
S.K. Chuang, L. Tian, L.J. Wei, and T.B. Dodson
Predicting Dental Implant Survival by Use of the Marginal Approach of the Semi-parametric Survival Methods for Clustered Observations
J. Dent. Res., December 1, 2002; 81(12): 851 - 855.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (30)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chuang, S.K.
Right arrow Articles by Dodson, T.B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chuang, S.K.
Right arrow Articles by Dodson, T.B.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
IADR Journals Advances in Dental Research ®
Journal of Dental Research ® Critical Reviews (1990-2004)