A 5-year Prospective validation of the Miller–McEntire Periodontal Prognostic Index

Abstract

Background:

This 5-year prospective survival analysis study aimed to examine the prognostic validity of a periodontal prognostic score specific for diseased molars: Miller–McEntire Periodontal Prognostic Index (MMPPI).

Materials and Methods:

One thousand and twenty-three molars were evaluated from 129 patients. The MMPPI scoring factors included age, smoking, diabetes, probing depth, mobility, molar type, and furcation involvement. MMPPI was computed as the sum of scores for all seven prognostic factors. Appropriate periodontal treatment and supportive periodontal therapy were provided. All patients were evaluated at baseline and annually posttreatment up to 5 years. Hazard risk ratios (HR) were computed for each prognostic factor, MMPPI scores assigned. The MMPPI score were then analyzed using Kaplan-Meier survival analyses.

Results:

A total of 31/1023 (0.3%) molars were extracted over the 5-year follow-up duration. Significant and positive hazard risk ratio (HR = 1.9) was noted for the total MMPPI score, validating its prognostic value for molar survival at 5 years prospectively. Kaplan–Meier survival analysis showed a significantly lower probability of molar survival with increasing MMPPI scores, where total score >8 showed worse survival probability over time. The hazard risk ratio was significant for individual prognostic factors: mobility (HR = 1.63), smoking (HR = 1.61), diabetes mellitus (DM) (HR = 1.4), molar type (1.97), and furcation involvement (2.22).

Conclusions:

The findings of the current study demonstrate significant prognostic validity of MMPPI scores for molar loss for 5 years, and a score >8 showed markedly worse molar survival probability in a well-maintained, university-based, prospective cohort. Mobility, smoking, DM, molar type, and furcation were component factors that were significant individual predictors.

INTRODUCTION

Periodontal prognosis is central to making effective therapeutic decisions. Multiple prognostic factors include modifiable and nonmodifiable factors.[1] Molar teeth are particularly vulnerable to periodontitis-related tooth loss, and a prognostic index specifically for periodontally diseased molars has been proposed.[2,3] Prognostic factors include probing depth, tooth mobility, furcation involvement, and smoking among others.[4,5] Miller et al.[2] evaluated seven prognostic factors that included age, smoking, diabetes, probing depth, molar type, mobility, and furcation involvement integrated into an easily assessed scoring system termed “Miller–McEntire Periodontal Prognostic Index (MMPPI)” initially assessed retrospectively.[2] Subsequent prospective studies validated the applicability of MMPPI index in both smoker and diabetic populations over 2 years.[6,7] In our previous study, the survival outcomes of 200 molar teeth for 2 years were assessed using this scoring system. Tooth-related prognostic factors such as furcation and mobility showed a significant and positive hazards ratio, predictive of molar tooth loss at 2 years.[6] At this time point, among all the assessed prognostic factors, smoking showed the highest impact with an HR of 3.35.[6] At the same time, diabetes mellitus (DM) showed an HR of 5.2 implying a five times greater risk for molar loss.[7] However, long-term outcomes in larger cohorts over a longer duration can further inform the validity of the MMPPI for clinical application. Hence, the present study was designed to prospectively assess the validity of the MMPPI in a large cohort over a period of 5 years.

MATERIALS AND METHODS

Study population and sampling

A prospective cohort study with a sample size of 129 patients and 1023 molar teeth was carried out for a follow-up duration of 5 years. Sampling was started in 2013 and follow-up was completed in 2019. A subsection of the patients from this cohort has been reported.[6,7] All study procedures were conducted after approval from the institutional ethics committee and have been described earlier, and written informed consent was obtained from all the 129 patients.[6,7] The sample size was computed assuming an expected event rate of tooth loss of 0.12 teeth/year, as reported for a periodontal maintenance group earlier, a planned follow-up period of 5 years, alpha = 0.05, and beta = 0.8.[8] At the end of the follow-up period, all patients exited by remaining in the regular periodontal maintenance program at the postgraduate periodontics clinics.

Inclusion and exclusion criteria

The inclusion criteria were patients with moderate-to-severe chronic periodontitis as described earlier,[6,7] and in brief, the exclusion criteria included restorations or prosthesis in molars or root canal treatment, and any known systemic or infectious disease which may affect periodontal prognosis.

Study procedures

All study procedures were aligned with the ethical principles of the Declaration of Helsinki, 2008. MMPPI scores for age, smoking, diabetes, molar type, probing depth, furcation involvement, and mobility were obtained for a total of 1023 molar teeth in 129 patients by two calibrated and experienced periodontists (S. S and G. K). The scoring criteria were as described earlier[6,7] and are represented in Table 1. The total of these 7 scores was computed as the MMPPI score. Periodontal treatment provided as described before[6,7] included nonsurgical periodontal therapy. Patients with residual pockets after nonsurgical therapy were considered for appropriate surgical intervention.[6,7] Periodontal maintenance therapy was provided biannually as described earlier.[6,7] Data from the 5-year examination regarding tooth survival were used to assess the validity of the prognostic index.

Table 1.

Miller–McEntire Periodontal Prognostic Index scoring criteria

Prognostic scoring parameter	Distribution of scores
Age (years)
<40	0
>40	1
Smoking
Nonsmoker	0
Occasional smoker	1
Smoking half pack/day	2
Smoking 1 pack/day	3
Smoking >1 pack/day	4
Diabetes (HbA1c)
<6	0
6-7	1
7.1-8	2
8.1-9	3
>9	4
Furcation involvement
No furcation involvement	0
1 furcation involved	1
2 furcations involved	2
3 furcations involved or through and through furcation	3
Molar type
Mandibular molar	0
Maxillary first molar	1
Maxillary second molar	2
Probing depth (mm)
<5	0
5-7	1
8-10	2
>10	3
Mobility
No mobility	0
Class I	1
Class II	2
Class III	3

HbA1c – Glycosylated hemoglobin

Data analysis

Molar survival analysis was done using Cox proportional hazards model, and hazard ratios were determined. Multivariate Cox proportional hazards models were applied for each of the seven MMPPI component factors, and a univariate model was applied for the total MMPPI score at baseline. All statistical analyses were performed in the R statistical environment (Foundation for Statistical Computing, Vienna, Austria, http://www.R-project.org/). The Cox proportional hazards[9] and KaplanMeier[10] analyses were done using the R package “survival.”[11] Kaplan-Meier survival times curves were plotted for four MMPPI score categories: 0–2, 2–4, 5–7, and >8 to visualize survival probabilities over time as associated with MMPPI scores, and a log-rank test was applied to compare survival curves. Data were right censored, assuming the event of tooth loss did not occur during the follow-up period.

RESULTS

One thousand and twenty-three teeth (129 patients) were assessed over 5 years. Descriptive statistics of the MMPPI component are presented in Table 2, and Figure 1 shows the frequency distribution of the MMPPI score. A total of 31 (0.3%) teeth were extracted over the 5-year follow-up duration, leading to a survival of 97%. The results of the Cox proportional hazards model and hazard ratio are summarized in Table 3. Significant HR of individual prognostic factors in the multivariate model was noted for mobility (HR = 1.63, P = 0.0), smoking (HR = 1.61, P = 0.02), molar type (HR = 1.97, P = 0.04), furcation involvement (HR = 2.22, P < 0.001), and DM (HR = 1.40, P = 0.04). Significant and positive hazard risk ratio (HR = 1.9, P < 0.001) was noted for the total MMPPI score. KaplanMeier survival times curves showed particularly lower survival probability over time for MMPPI score >8 [Figure 2] and a highly significant difference between the four MMPPI categories (Chi-square statistic = 230, P < 0.001).

Table 2.

Distribution of prognostic variables and Miller- McEntire Periodontal Prognostic Index score at baseline

Prognostic scoring parameter	Distribution of scores at baseline
Age (years)
<40: 69 patients/560 teeth	0
>40: 60 patients/474 teeth	1
Smoking
Nonsmoker: 104 patients/792 teeth	0
Occasional smoker: 22 patients/175 teeth	1
Smoking half pack/day: 1 patient/7 teeth	2
Smoking 1 pack/day: 2 patients/15 teeth	3
Smoking >1 pack/day: 0 patients/0 teeth	4
Diabetes (HbA1c)
<6: 102 patients/814 teeth	0
6-7: 4 patients/31 teeth	1
7.1-8: 3 patients/24 teeth	2
8.1-9: 20 patients/157 teeth	3
>9: 0 patients/0 teeth	4
Furcation involvement
No furcation involvement: 845 teeth	0
1 furcation involved: 98 teeth	1
2 furcations involved: 60 teeth	2
3 furcations involved or through and through: 20 teeth	3
Molar type
Mandibular molar: 511 teeth	0
Maxillary first molar: 256 teeth	1
Maxillary second molar: 256 teeth	2
Probing depth (mm)
<5: 355 teeth	0
5-7: 492 teeth	1
8-10: 155 teeth	2
>10: 21 teeth	3
Mobility
No mobility: 881 teeth	0
Class I: 113 teeth	1
Class II: 11 teeth	2
Class III: 18 teeth	3

HbA1c - Glycosylated hemoglobin

Figure 1.

Histogram depicting frequency distribution of Miller–McEntire Periodontal Prognostic Index score values (MMPPI).

Table 3.

Cox proportional hazards model results for molar survival at 5 years

Prognostic factor	Parameter estimate	Hazard ratio (95% CI)	SE	Z	P
Age*	0.62	1.85 (0.78-4.38)	0.44	1.40	0.16
Mobility*	0.49	1.63 (1.01-2.63)	0.24	2.00	0.04
Smoking*	0.48	1.61 (1.07-2.43)	0.21	2.29	0.02
Diabetes*	0.34	1.40 (1.03-2.09)	0.21	1.63	0.04
Molar type*	0.68	1.97 (1.23-3.15)	0.24	2.80	0.01
Probing depth*	0.57	1.77 (1-3.147)	0.30	1.93	0.04
Furcation*	0.80	2.22 (1.47-3.54)	0.21	3.79	<0.001
MMPPI score#	0.644	1.9 (1.65-2.20)	0.074	8.7	<0.001

*Multivariable model with seven prognostic factor scores as predictor: Likelihood ratio test=81.8 on 7 df; P<0.05; ^#Univariate model with MMPPI score as predictor: Likelihood ratio test=76.8 on 1 df; P<0.001 (significant P values in bold). SE - Standard error; Z - Wald statistic value; P - Probability value; MMPPI - Miller–McEntire Periodontal Prognostic Index; CI - Confidence interval

Figure 2.

KaplanMeier survival times curves for Miller–McEntire Periodontal Prognostic Index score categories (MMPPI).

DISCUSSION

Overall, the present data confirmed the initial findings of the earlier short-term prospective studies,[6,7] and the prognostic value of the MMPPI score was prospectively validated in a large cohort that included subjects with both DM and habit of smoking. The earlier samples had reported the value of component scores for 2-year periods in specific populations with DM and smoking.[6,7]

Here, a highly significant and positive hazard risk ratio (HR = 1.9) was noted for the total MMPPI score, validating its prognostic value for molar survival at 5 years prospectively. Thus, throughout 5 years, the risk of tooth loss was approximately doubled for each unit increase in the MMPPI score. The tooth loss (31 teeth) was higher than absolute numbers that were noted in the earlier samples followed for 2 years.[6,7] Miller et al.[2] had noted a mean annual tooth loss of 1.7 molars in a retrospective cohort assessed at least after 15 years of initial therapy. Five-year data from this well-supervised university cohort indicated a lower rate of tooth loss, which is likely to have cumulative incidence over time, and comparable rates have been reported earlier in periodontal maintenance populations followed over similar durations.[12]

A very important finding from these data is that the Kaplan–Meier survival analysis showed a significantly lower probability of molar survival with increasing MMPPI scores [Figure 2], and particularly, a total score >8 showed a much worse survival probability over time, validating the clinical applicability of the score. This finding can have a valuable clinical translation in objectively quantifying risk during clinical examination, where only the molars with scores over 8 may be categorized as having a guarded prognosis. Routine application of such a validated prognostic score threshold can aid in avoiding unnecessary extractions, by the more precise risk categorization afforded by MMPPI, thus strongly supporting its use in clinical decision-making. Thus, this objective, evidence-based periodic scoring can aid clinicians in obtaining better clinical outcomes and also reinforce and motivate patients for lifestyle modification.[13] Here, MMPPI scores were applied to each molar assessed, as the study was designed for epidemiological index validation. In practice, the worst affected molar that the clinician is planning to keep may be used as an index, as described earlier, in order to facilitate rapid clinical application and patient acceptance.[13]

The multivariate Cox proportional hazards models demonstrated individual component prognostic factors’ contribution to the risk of molar loss similar to earlier data.[6,7,13,14,15,16] The 2-year data had demonstrated significant positive hazard risk ratios for smoking, DM, mobility, and furcation involvement[6,7] In the present analysis, largely in agreement with the previous reports,[4,5,6,7,14,15] the 5-year hazard risk ratio for tooth loss was found significant for mobility, smoking, DM, molar type, and furcation involvement. Similar to the earlier report,[6] a significant hazard ratio for DM was noted. This was noted despite the gradual improvement of diabetic control over time due to medical supervision and regular supportive periodontal care in this prospective cohort placed in an academic setting. In this study population, at baseline, twenty patients were found to have HbA1c levels of 8.1–9 (score 3), but by 2 years, none retained this poor level of glycemic control. Molar loss in this group was largely concentrated in the early observation period. Others have reported that well-controlled diabetics do not have an increased risk of periodontitis, and it is fathomable that improvement of metabolic control could lower the HR conferred by a DM diagnosis.[16] Among the component factors, as found earlier, the tooth-level prognostic factor furcation involvement conferred the highest risk. This finding may be correlative to the fact that furcation areas may serve as sites for continued disease progression, owing to the challenge of inadequate debridement in these areas and also challenges in achieving successful regeneration in teeth with multiple furcation involvements. Previous studies[17,18] support a risk for disease progression in furcation sites despite periodontal therapy. Notably, the absolute value of the hazard ratio in the present data (2.2) was much lower than that reported at 2 years for a smaller cohort with smokers (7.3),[7] which could be owing to variation in smoking dose over time. Smoking can contribute to worsening of furcation involvement in molar teeth, and adversely impact the outcomes of regenerative surgery, thus an interaction effect is likely.[19] As molars with furcation involvement have shown an increased risk of mortality over time, especially over periods of 10 years or more,[20] longer periods of assessment can be further informative about how smoking and furcation involvement impact the risk of tooth loss over time. The HR for the molar type was closely in agreement with the retrospective data from Miller et al.[2] (1.20) indicating a higher risk for loss of maxillary molars as compared to mandibular molars. While the major strength of the study is the 5-year duration follow-up, a limitation of the current application of this index is that the need for the type of therapy, need for surgery, need for re-treatment, and time points of surgical intervention as outcomes have not been addressed. Furthermore, as the patients were under a stringent care regime in a university setting, the generalizability of the index to real-world clinical populations needs to be ascertained. In addition, other prognostic factors such as bone loss/age ratio, number of teeth lost, and comparison with other prognostic tools such as periodontal risk assessment and therapy-resistant periodontitis assessment warrant investigation in future.[21]

CONCLUSIONS

The current study demonstrated that a high MMPPI score >8 predicted significantly worse molar mortality and each unit increase in MMPPI score translated to 1.9 times increase in risk of molar loss over 5 years in a well-supervised, university-based cohort. As MMPPI component factors and their thresholds are amenable to rapid clinical recording, these data support its large-scale application for more precise clinical decision-making.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.