Supportive Periodontal Therapy Clinical Examination Data

@article{PhysioNet-supportive-periodontal-therapy-1.0.0,
  author = {Ramseier, Christoph Andreas},
  title = {{Supportive Periodontal Therapy Clinical Examination Data}},
  journal = {{PhysioNet}},
  year = {2026},
  month = oct,
  note = {Version 1.0.0},
  doi = {10.13026/cn9b-2632},
  url = {https://doi.org/10.13026/cn9b-2632}
}

@article{pollard_physionet_2026, title = {{PhysioNet} as a global platform for biomedical research}, volume = {1}, issn = {3005-0693}, url = {https://doi.org/10.1038/s44360-026-00096-z}, doi = {10.1038/s44360-026-00096-z}, abstract = {Twenty-five years after its first public repository was established, the PhysioNet resource for biomedical and clinical data has enabled tens of thousands of research papers. Further efforts are ongoing to advance secure data sharing, community engagement and scalable research infrastructure.}, number = {8}, journal = {Nature Health}, author = {Pollard, Tom and Moody, Benjamin E. and Lehman, Li-wei H. and Gow, Brian J. and Fernandes, Chrystinne and Xie, Chen and Johnson, Alistair and Mark, Roger G. and Heldt, Thomas}, month = aug, year = {2026}, pages = {792--795},}

Abstract

This dataset contains longitudinal clinical data from 883 patients who received supportive periodontal therapy (SPT) at the Medi School of Dental Hygiene, Bern, Switzerland, 1985-2011. It covers 11,842 SPT visits, each recording the teeth present, the sites bleeding on probing, the sites in each residual probing depth (PPD) category, and the time since the preceding visit. Patient-level variables comprise age, sex, smoking status, medical history, disease severity at the initial examination, and findings before and after active therapy.

The data were collected during routine care and assembled to address questions maintenance protocols answer only loosely, beginning with how long the interval between visits should be. They supported analyses of patient compliance with scheduled visits and of bleeding on probing in relation to smoking status, and the derivation and validation of an algorithm, now in clinical use, that computes the interval to the next visit from a patient's residual probing depth profile.

Probing depths were recorded at six sites per tooth and bleeding on probing at four. Sites of 0 to 3 mm were not recorded and depths of 8 mm or more were not distinguished from 7 mm. Because intervals were assigned from bleeding scores and disease severity rather than at random, comparisons of interval length with outcome are confounded by indication. Every manual edit made during data preparation is documented visit by visit in a separate file, and scripts in R and Python reproduce the algorithm and the published stability thresholds from these files.

Background

Periodontal disease is a chronic inflammatory condition. Active periodontal therapy can reduce the inflammatory burden and arrest progression, but the chronic nature of the disease makes lifelong supportive periodontal therapy essential to maintain the treatment outcome [4]. A cohort followed over thirty years showed that patients attending regular supportive care preserved periodontal stability and lost few teeth, whereas patients who did not attend showed deteriorating conditions, increasing probing depths and more frequent tooth loss [5,6]. A systematic review and meta-analysis found the risk of tooth loss to be reduced in patients who complied more closely with supportive periodontal therapy [7].

How often a patient should return is less well established than the fact that they should. Current practice determines the interval mainly from the percentage of bleeding on probing recorded at each visit: patients with a mean bleeding score at or below 20 per cent are regarded as periodontally stable and may be scheduled at longer intervals, while patients with higher scores are scheduled earlier [8]. Structured risk assessment tools combine bleeding scores with further patient-level parameters, including the number of residual sites with probing depths of 5 mm or more [9,10]. Such tools predict disease progression and tooth loss, but they do not yield a recommended length of the interval between visits [11].

Residual probing depth itself carries prognostic information. A retrospective site-based analysis found periodontal stability in patients with residual probing depths of 5 mm or lower, while probing depths of 6 mm or more were positively associated with disease progression and subsequent tooth loss [12]. Bleeding on probing is moreover an unreliable indicator in smokers, who do not show elevated bleeding scores concomitantly with increased residual probing depths [2]. Monitoring the full profile of residual probing depths at every visit may therefore add information that bleeding scores alone cannot supply.

The data published here were collected over 26 years at a single dental hygiene school and were used to address that gap. They supported an analysis of patient compliance with scheduled visits and its association with smoking [1], an analysis of bleeding on probing in relation to smoking status [2], and the derivation and validation of an algorithm that computes the interval between supportive periodontal therapy visits from the patient's residual probing depth profile [3]. The algorithm is in clinical use and is available as an online tool [13].

Longitudinal periodontal data covering more than two decades of maintenance care are rare, and few clinical prediction tools used in periodontology can be recomputed from their source data. This dataset is shared so that the published analyses can be verified independently, and so that the question it addresses — how to individualise the timing of maintenance care — can be pursued with methods that were not available when the data were collected.

Methods

The dataset was derived from the clinical records of 883 patients who received supportive periodontal therapy at the Medi School of Dental Hygiene (MSDH), Bern, Switzerland, between 1985 and 2011. Patients were included if they were aged 20 years or older, had completed active periodontal therapy, had attended at least two supportive periodontal therapy visits within one year, and had complete demographic and clinical records.

Upon entry at the MSDH, all patients were fully examined and diagnosed. Periodontally healthy patients received prophylaxis; periodontally diseased patients were treated by non-surgical periodontal therapy carried out by dental hygiene students, with periodontal surgery for pocket reduction where indicated. Following prophylaxis or active therapy, all patients entered the MSDH supportive periodontal therapy programme.

Probing depths were recorded at six sites per tooth using a calibrated periodontal probe. Measurements were taken by dental hygiene students and re-measured by experienced clinical instructors; student recordings were corrected where they differed from the instructor's. Sites with probing depths of 0 to 3 mm were not recorded, and depths of 8 mm or more were not recorded separately from 7 mm. Bleeding on probing was recorded at four sites per tooth. Clinical data collected at each visit comprise the number of teeth, the number of sites bleeding on probing, and the number of sites at each residual probing depth category.

The interval to the next visit was determined at the end of active therapy and at each subsequent visit from the percentage of bleeding on probing and the severity of periodontal disease. Where bleeding on probing was below 20 per cent, the previously determined interval was increased by one to two months up to a maximum of twelve; where it was 20 per cent or above, the interval was decreased by one to two months down to a minimum of three.

Severity of periodontal disease at the initial examination was categorised as I) no or mild disease, where no site exceeded a probing depth of 4 mm, III) advanced disease, where at least ten sites showed probing depths of 6 mm or more, and II) moderate disease for all remaining patients. This category was recorded in the original study files for 445 patients and has been reconstructed from the baseline site counts for the remaining 438; the diagnosis_source variable marks which value came from where. The reconstruction agrees with every one of the 445 recorded values.

A subsample of 445 patients was used in reference [2] and, in reference [3], both in the validation of the interval algorithm and in the mixed-effects model of Table 2. It comprises those patients who were systemically healthy at the initial examination and who attended supportive periodontal therapy for at least five years.

De-identification. Patient identifiers are sequential study numbers that cannot be linked back to clinical records. All calendar dates present in the working data — dates of the initial examination, of the end of active therapy and of every visit — were removed before publication; only the year of each visit is retained. No date shifting was applied, because no date other than the year is included. Ages above 89 years are aggregated to 90 in accordance with the HIPAA Safe Harbor standard.

Manual edits during data preparation. Sixty-three of the 11,842 visits received manual attention when these data were first prepared, and this was documented at the time. In 47 visits no probing depth measurement existed and values were inserted, in 41 cases by carrying forward the preceding visit; those values have been removed and the visits are flagged by the imputed variable. In 16 visits a measurement existed but was judged implausible and was replaced; the replacement is retained, because the published analyses rest on it, and the visits are flagged by the corrected variable. Both groups are listed individually, with their original and their used values, in 04_data_corrections.csv. No visits were removed.

The algorithm that computes the interval between visits was derived from empirically determined thresholds below which no increase of residual probing depth was observed between consecutive visits, separately for residual probing depths of 4 mm or more, 5 mm or more and 6 mm or more, and for interval lengths of 3, 4, 6 and 12 months [3]. Residual probing depths of 7 mm or more were not used, because depths of 8 mm or more were not recorded and the category therefore did not qualify as a threshold variable. The full algorithm is provided as an R script and as a Python script with this dataset.

Data Description

The dataset consists of four comma-separated value files and a data dictionary. All files are encoded in UTF-8 without a byte order mark and formatted according to RFC 4180. Missing values are empty fields.

01_initial_periodontal_therapy_data.csv — 883 rows, 22 variables. One row per patient: demographics, medical history, smoking status, disease severity, clinical findings before and after active periodontal therapy, and summary measures of the supportive phase.

02_supportive_periodontal_therapy_data.csv — 11,842 rows, 15 variables. One row per supportive periodontal therapy visit: clinical findings, assigned and algorithm-based intervals, and the time elapsed since the preceding visit.

03_supplementary_data.csv — 11,842 rows, 44 variables. One row per visit, containing variables derived from the primary data: cumulative site counts, percentages, the intermediate steps of the interval algorithm, and adherence to the computed interval.

04_data_corrections.csv — 63 rows, 12 variables. Every visit whose probing depth values differ from the source records, with the original value, the value used, and the nature of the edit.

dictionary.csv — description of all 93 variables of the four data files.

Six analysis scripts accompany the data, three in R and three in Python: 01_compute_spt_algorithm, 02_linear_mixed_effects_model and 03_reproduce_figure_2. Each pair was written against the same specification and the two implementations agree. README.md summarises the dataset, the file structure and the accompanying code.

Files 02 and 03 are joined on pat_id and spt_id; both join to file 01 on pat_id. Probing depth was recorded at six sites per tooth and bleeding on probing at four, so probing depth percentages use n_teeth_spt * 6 as the denominator and bleeding on probing percentages use n_teeth_spt * 4.

Patients at the initial examination (n = 883):

Sex: 489 female (55.4 per cent), 394 male (44.6 per cent). Smoking status: 394 non-smokers (44.6 per cent), 222 former smokers (25.1 per cent), 267 smokers (30.2 per cent). Medical history: 788 systemically healthy (89.2 per cent), 49 on anticoagulants, 23 on antidepressants, 7 miscellaneous, 6 on antibiotic prophylaxis, 3 with an infectious disease, 2 on antiepileptic drugs, 5 not recorded. Severity of periodontal disease: 147 category I (16.6 per cent), 468 category II (53.0 per cent), 268 category III (30.4 per cent).

Supportive periodontal therapy visits (n = 11,842):

The data cover the years 1985 to 2011. Patients attended a median of 9 visits, ranging from 2 to 70, over a median of 6.0 years and up to 25.7 years. The algorithm-based interval was 3 months at 3,003 visits, 4 months at 955, 6 months at 3,646 and 12 months at 4,191. Comparing the actual interval with the computed one, patients attended earlier than computed at 4,454 visits, at the computed interval at 2,338, and later at 5,003.

Counts of 11,795 rather than 11,842 reflect the 47 visits for which no probing depth measurement existed and whose inserted values have been removed. Bleeding on probing is missing at the 26 visits where it was not recorded.

Usage Notes

Reproducing the original analyses. Three published analyses can be reproduced from these files. They differ in the study population, which is identified by variables in file 01: n_spt_visits gives the number of supportive periodontal therapy visits, spt_duration_years the time from the first to the last recorded visit, subsample_5years marks the 445 systemically healthy patients who attended for at least five years (equivalent to med_history == "healthy" and spt_duration_years >= 5), and diagnosis together with diagnosis_source gives the severity category at the initial examination and its provenance.

Reference [1], compliance with scheduled visits, used all 883 patients. It draws on compliance (file 03) and on smoking, gender and age_baseline (file 01).

Reference [2], bleeding on probing and smoking status, used the 445 patients with subsample_5years == 1. It draws on n_bop_pos_spt and the residual probing depth counts (file 02) and on smoking and diagnosis (file 01).

Reference [3], time between recall visits and residual probing depths, used all 883 patients for Table 1 and Figure 2. Table 2 and the validation of the algorithm used the 445 patients with subsample_5years == 1.

Accompanying code. Three scripts are provided, each in R and in Python. The two implementations of each were written against the same specification and agree. 01_compute_spt_algorithm reproduces every derived variable of file 03 from the primary data, including the interval algorithm, and requires only base R or pandas and numpy. 02_linear_mixed_effects_model reconstructs the linear mixed-effects model of Table 2 in reference [3]; the R version requires lme4 and lmerTest. 03_reproduce_figure_2 reproduces the probing depth stability thresholds of Figure 2 in reference [3] and reports a sensitivity analysis.

The source code written for reference [3] is no longer available. Script 01 reproduces all 11,842 rows of the derived data exactly, and script 03 reproduces all twenty published stability thresholds exactly; both verify themselves against the published files when run and print the result. Script 02 is a reconstruction of the published model specification. It reproduces the published estimates, standard errors and confidence intervals closely, and its header documents what the reconstruction rests on, which population Table 2 was fitted on, and how the F column of that table relates to the intervals beside it.

Reuse potential. The dataset supports research into the individualisation of maintenance care: the validation or development of algorithms for scheduling supportive periodontal therapy, the modelling of periodontal stability as a function of the time between visits, and the study of patient-level determinants of attendance. It is also suitable for teaching, as a compact longitudinal clinical dataset with a transparent derived outcome.

Known limitations. The intervals between visits were not assigned at random. They were determined from bleeding on probing and disease severity, so that patients with more severe disease systematically received shorter intervals. Any comparison of interval length with clinical outcome is therefore confounded by the rule that generated the assignment. Adherence compounds this: patients who consistently attend earlier than scheduled are likely to differ in oral hygiene, smoking and health behaviour generally. A retrospective cohort cannot separate the effect of the interval from the characteristics of the patients who keep it.

All data were collected at one dental hygiene school in Switzerland between 1985 and 2011, under one standard of care; generalisability to other populations, care settings and periods is untested. Plaque scores, clinical attachment levels and radiographic findings were not collected, and reasons for tooth loss were not recorded. Probing depths of 0 to 3 mm were not recorded, and depths of 8 mm or more were not distinguished from 7 mm. Smoking status was recorded at the initial examination and was not updated during follow-up. The compliance measure is provided as computed for reference [1], where its definition is given; it cannot be recomputed exactly from the files published here.

Reading Table 2 of reference [3]. Three points matter to anyone recomputing that analysis, and all three are documented in the header of the accompanying script.

First, the model was fitted on the 445-patient subsample, not on all 883. The covariate for disease severity was recorded only for those patients, so a complete-case fit is confined to them, and the scale of the published estimates confirms it: the outcome is rank transformed, so every coefficient is proportional to the number of observations in the fit. The subsample fit reproduces the published coefficient for the residual probing depth at the visit to within one per cent; the full-cohort fit is twenty-eight per cent too large.

Second, the F values reported in the table are sequential (type I) tests, while the confidence intervals in the same table are marginal. For bleeding on probing the two diverge sharply: the interval spans zero while F = 283.15. Both are correct and answer different questions. Bleeding on probing and the residual probing depth at the same visit are correlated (r = 0.39); entered first in the sequence, bleeding on probing carries the variance they share, and adjusted for probing depth its independent contribution is negligible. The script reports both decompositions, and the marginal tests are the ones to build on.

Third, two F values in the printed table carry a decimal error: periodontal disease is shown as 0.06 and smoking status as 0.12, against 0.6330 and 0.0123 in the statistician's working document, which the reconstruction supports. Both terms are far from significance in every version.

Release Notes

Version 1.0.0

This is the first release of the dataset.

During the original preparation of these data, 63 of the 11,842 supportive periodontal therapy visits received manual attention, and this was documented at the time. In 47 visits no probing depth measurement existed and values were inserted, in 41 cases by carrying forward the preceding visit. In 16 visits a measurement existed but was judged implausible and was replaced.

The inserted values are not part of this release: for those 47 visits the probing depth counts and everything derived from them are missing, while the tooth count and the bleeding on probing count, which were measured, are retained. The replacements in the remaining 16 visits are included, because the published analyses rest on them. Both groups are flagged by the imputed and corrected variables, and 04_data_corrections.csv lists every one of the 63 visits with its original and its used values, so that the state on which the published analyses rest can be restored exactly.

All calendar dates have been removed; only the year of each visit is retained. Ages above 89 years have been aggregated to 90 in accordance with the HIPAA Safe Harbor standard.

Three analysis scripts accompany the data, each provided in R and in Python.

Ethics

This dataset was derived from the clinical records of 883 patients who received supportive periodontal therapy at the Medi School of Dental Hygiene, Bern, Switzerland, between 1985 and 2011. The data were collected in the course of routine clinical care and anonymised before analysis. Permission to conduct the original studies was granted by the Medi School of Dental Hygiene in 2011.

The publication of this dataset was submitted to the Cantonal Ethics Committee of Bern, Switzerland, for a clarification of responsibility (BASEC Req-2025-00339). In its decision of 26 March 2025, the committee determined that the project does not fall within the scope of the Swiss Human Research Act and that approval by an ethics committee is therefore not required. The committee further stated that it is not responsible for the evaluation of the project and that data protection must be guaranteed.

No personally identifiable information is contained in the dataset. Patient identifiers are sequential study numbers that cannot be linked back to clinical records. All calendar dates have been removed; only the year of each visit is retained. Ages above 89 years are aggregated to 90 in accordance with the HIPAA Safe Harbor standard.

The risks associated with the reuse of this dataset are minimal. Its publication is intended to support research into individualised periodontal maintenance care and the development of predictive models for periodontal disease management.

Acknowledgements

This dataset and the associated study were supported by the Swiss National Programme to Stop Smoking and a research grant from the Swiss Dental Association (SSO, Grant No. 306-18). The authors thank Dr Herbert Hofstetter, Dr Petra Hofmänner and the administrative staff of the Medi School of Dental Hygiene, Bern, Switzerland, for their invaluable contributions. Special thanks are due to Dr Salome Kobrehel and Dr Petra Staub for their assistance in extracting data from patient records.

Conflicts of Interest

The authors declare no conflicts of interest related to the creation, analysis, or sharing of this dataset.

References

1. Ramseier CA, Kobrehel S, Staub P, Sculean A, Lang NP, Salvi GE. Compliance of cigarette smokers with scheduled visits for supportive periodontal therapy. J Clin Periodontol. 2014;41(5):473-480. doi:10.1111/jcpe.12242
2. Ramseier CA, Mirra D, Schütz C, Sculean A, Lang NP, Walter C, et al. Bleeding on probing as it relates to smoking status in patients enrolled in supportive periodontal therapy for at least 5 years. J Clin Periodontol. 2015;42(2):150-159. doi:10.1111/jcpe.12344
3. Ramseier CA, Nydegger M, Walter C, Fischer G, Sculean A, Lang NP, et al. Time between recall visits and residual probing depths predict long-term stability in patients enrolled in supportive periodontal therapy. J Clin Periodontol. 2019;46(2):218-230. doi:10.1111/jcpe.13041
4. Lindhe J, Socransky SS, Nyman S, Haffajee A, Westfelt E. "Critical probing depths" in periodontal therapy. J Clin Periodontol. 1982;9(4):323-336. doi:10.1111/j.1600-051X.1982.tb02099.x
5. Axelsson P, Nyström B, Lindhe J. The long-term effect of a plaque control programme on tooth mortality, caries and periodontal disease in adults. Results after 30 years of maintenance. J Clin Periodontol. 2004;31(9):749-757. doi:10.1111/j.1600-051X.2004.00563.x
6. Axelsson P, Lindhe J. The significance of maintenance care in the treatment of periodontal disease. J Clin Periodontol. 1981;8(4):281-294. doi:10.1111/j.1600-051X.1981.tb02039.x
7. Lee CT, Huang HY, Sun TC, Karimbux N. Impact of patient compliance on tooth loss during supportive periodontal therapy: a systematic review and meta-analysis. J Dent Res. 2015;94(6):777-786. doi:10.1177/0022034515578910
8. Lang NP, Adler R, Joss A, Nyman S. Absence of bleeding on probing. An indicator of periodontal stability. J Clin Periodontol. 1990;17(10):714-721. doi:10.1111/j.1600-051X.1990.tb01059.x
9. Lang NP, Tonetti MS. Periodontal risk assessment (PRA) for patients in supportive periodontal therapy (SPT). Oral Health Prev Dent. 2003;1(1):7-16.
10. Matuliene G, Studer R, Lang NP, Schmidlin K, Pjetursson BE, Salvi GE, et al. Significance of periodontal risk assessment in the recurrence of periodontitis and tooth loss. J Clin Periodontol. 2010;37(2):191-199. doi:10.1111/j.1600-051X.2009.01508.x
11. Lang NP, Suvan JE, Tonetti MS. Risk factor assessment tools for the prevention of periodontitis progression: a systematic review. J Clin Periodontol. 2015;42(Suppl 16):S59-S70. doi:10.1111/jcpe.12350
12. Matuliene G, Pjetursson BE, Salvi GE, Schmidlin K, Brägger U, Zwahlen M, et al. Influence of residual pockets on progression of periodontitis and tooth loss: results after 11 years of maintenance. J Clin Periodontol. 2008;35(8):685-695. doi:10.1111/j.1600-051X.2008.01245.x
13. Ramseier CA. Supportive periodontal therapy interval tool [Internet]. Bern: perio-tools.com [cited 2026 Sep 12]. Available from: https://www.perio-tools.com/spt