Effect of electronic reminders on patients' compliance during clear aligner treatment: an interrupted time series study

Abstract

Patient compliance is relevant to achieving therapeutic goals during clear aligner therapy (CAT). The aim of this study was to evaluate the efficacy of remote electronic (e-)reminders and e-feedback on compliance during CAT using an interrupted time series (ITS) analysis. We used routinely collected mobile application data from a German healthtech company (PlusDental, Berlin). Our primary outcome was self-reported compliance (aligner wear time min. 22 h on 75% of their aligners were classified as fully compliant, min. 22 h on 50–74.9% of their aligners: fairly compliant; min. 22 h on < 50% of their aligners: poorly compliant). E-reminders and e-feedback were introduced in the 1st quarter of 2020. Compliance was assessed at semi-monthly intervals from June-December 2019 (n = 1899) and June-December 2020 (n = 5486), resulting in a pre- and post-intervention group. ITS and segmented regression modelling were used to estimate the effect on the change in levels and trends of poor compliance. Pre-intervention, poor compliance was at 24.47% (95% CI: 22.59% to 26.46%). After the introduction of e-reminders and e-feedback (i.e., post-intervention), the percentage of poorly compliant patients decreased substantially, levelling off at 9.32% (95% CI: 8.31% to 10.45%). E-reminders and e-feedback were effective for increasing compliance in CAT patients.
Clinical Significance: Orthodontists and dentists may consider digital monitoring and e-reminders to improve compliance and increase treatment success.

Full text

1
Vol.:(0123456789)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports
Eect of electronic reminders
on patients’ compliance
during clear aligner treatment:
an interrupted time series study
Lan Huong Timm1,4,5*, Gasser Farrag1,5, Daniel Wolf1, Martin Baxmann2 & Falk Schwendicke3
Patient compliance is relevant to achieving therapeutic goals during clear aligner therapy (CAT). The
aim of this study was to evaluate the ecacy of remote electronic (e-)reminders and e-feedback on
compliance during CAT using an interrupted time series (ITS) analysis. We used routinely collected
mobile application data from a German healthtech company (PlusDental, Berlin). Our primary
outcome was self-reported compliance (aligner wear time min. 22 h on 75% of their aligners were
classied as fully compliant, min. 22 h on 50–74.9% of their aligners: fairly compliant; min. 22 h
on < 50% of their aligners: poorly compliant). E-reminders and e-feedback were introduced in the
1st quarter of 2020. Compliance was assessed at semi-monthly intervals from June-December 2019
(n = 1899) and June-December 2020 (n = 5486), resulting in a pre- and post-intervention group. ITS and
segmented regression modelling were used to estimate the eect on the change in levels and trends
of poor compliance. Pre-intervention, poor compliance was at 24.47% (95% CI: 22.59% to 26.46%).
After the introduction of e-reminders and e-feedback (i.e., post-intervention), the percentage of
poorly compliant patients decreased substantially, levelling o at 9.32% (95% CI: 8.31% to 10.45%).
E-reminders and e-feedback were eective for increasing compliance in CAT patients.
Clinical Signicance: Orthodontists and dentists may consider digital monitoring and e-reminders to
improve compliance and increase treatment success.
Patient compliance during orthodontic therapy (e.g., towards wearing a removable appliance, attending regular
re-evaluation and adaptation visits, adhering to specic oral hygiene requirements) has been shown highly
relevant to achieving therapeutic goals and reducing adverse eects1–4. Both chairside approaches (oentimes
involving written information, videos, or teaching aids, and demanding signicant chairside time, i.e. generat-
ing considerable costs)5 and remote interventions (e.g. telephone, text, or online reminders, which are scalable
and come at high accessibility) have been suggested to improve compliance5–8; the latter have been proven to
improve patient compliance in healthcare in general9–11. Considering the widespread use of smartphones globally,
remote interventions through messages and mobile applications oer a viable, low-cost, and equitable strategy
to improve compliance4,11.
Studies have shown that mobile short messaging could have a positive impact on short-term behavioural
outcomes12 and that active reminders (weekly text messaging) for patients undergoing orthodontic treatment
could improve oral health13. Weekly reminders were also found to improve the memorability of the information
given by orthodontists in the oce, increasing compliance3.
During clear aligner treatment (CAT), electronic messaging and self-management or regular notications
by the treating dentist using mobile applications may improve patients’ understanding of their therapy, their
compliance, and the resulting outcomes. e ecacy of such remote electronic reminders on compliance during
CAT, however, remains unclear.
Using the interrupted time-series (ITS) approach, we aimed to measure the impact of active electronic (e-)
reminders and automatic e-feedback on patient compliance during CAT. e objectives of the present study were
OPEN
1Sunshine Smile GmbH, Windscheidstraße 18, 10627 Berlin, Germany. 2Orthodentix, Arnoldstrasse 13 b,
47906 Kempen, Germany. 3Department of Oral Diagnostics, Digital Health and Health Services Research, Charité
– Universitätsmedizin Berlin, Aßmannshauser Straße 4-6, 14197 Berlin, Germany. 4DrSmile - DZK Deutsche
Zahnklinik GmbH, Königsallee 92a, 40212 Düsseldorf, Germany. 5These authors contributed equally: Lan Huong
Timm and Gasser Farrag. *email: lan.timm@drsmile-group.com
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
Vol:.(1234567890)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
(1) to analyse patient compliance before and aer the introduction of an e-reminder system during CAT, and
(2) to evaluate whether the eects of this reminder system varied by gender and age.
Materials and methods
Study design and sample. An ITS analysis was conducted to evaluate the eect of using e-reminders and
e-feedback on patient compliance during CAT. Data were collected by PlusDental, a brand of the Sunshine Smile
GmbH (Berlin, Germany), during the course of routine treatment of patients, and were anonymized for research
use. e study was conducted according to the guidelines of the Declaration of Helsinki. e data were collected
retrospectively as a part of the treatment and anonymized for research use, which according to the Berlin State
Hospital Act (Landeskrankenhausgesetz Berlin) and the recommendations of the Datenschutz und IT-Sicher-
heit im Gesundheitswesen (DIG) task force of the German Association for Medical Informatics, Biometry, and
Epidemiology (GMDS) does not require approval from an ethics committee.
All patients were instructed to report each aligner change as well as the daily aligner wearing time using the
app-based questionnaire as described in detail by Timm etal.14. e aligner change interval depended on the
prescribed wear protocol, either aer every 7days or aer 14days of wear. Hence, reporting of changing aligners
and average wear time were provided every 1–2weeks.
We classied patients into fully, fairly and poorly compliant. Aligner wear time of ≥ 22h with ≥ 75% of align-
ers and consistent use of the mobile application for aligner check-in were classied as full compliance. Patients
with inconsistent app use were classied as either fairly compliant or poorly compliant based on aligner wear
time: Aligner wear time of ≥ 22h with 50–74.9% of aligners was classied as fair compliance, and aligner wear
time of ≥ 22h with only < 50% of aligners was classied as poor compliance. us, patients who provided no or
inadequate information about their compliance would be classied as poorly compliant.
e implementation of e-reminders and automatic e-feedback was our intervention. e implementation
occurred at the beginning of 2020, i.e., as of the 1st quarter of 2020 onwards these reminders were used (see
below). e reporting of the study follows the STrengthening the Reporting of OBservational studies in Epide-
miology (STROBE) checklist; the employed methodology is in accordance with the World Medical Association
Declaration of Helsinki15.
Data of all patients aged 18–64years that (based on one intraoral scan) nished their treatment successfully
with the so-called 1–1-2 system14 (see subsection: "Intervention and data collection") during the second half of
2019, i.e., never received e-reminders or e-feedback during their therapy (n = 1899 patients) and the second half
of 2020, i.e., always received e-reminders and e-feedback (n = 5486 patients) were available. e introduction of
e-reminders and e-feedback was the only dierence between the two versions of the mobile application. A com-
prehensive sample was drawn (i.e., no random sampling, etc.). e patients showed the following characteristics
prior to therapy: Malocclusion in the anterior and/or premolar region to be treated with CAT; adults (> 18years)
with a permanent dentition; absence of active periodontal disease or local and/or systemic conditions that can
aect bone metabolisms; no extractions being required for the orthodontic treatment (Fig.1).
Intervention and data collection. A new e-reminder and e-feedback had been implemented and
deployed to the PlusDental mobile application in the rst quarter of 2020, reminding patients towards the 22h
aligner wearing time for ensuring tooth movement and instructing them to remove aligners only for eating,
drinking, and oral hygiene. If patients recorded a wearing time of less than 22h per day (they were actively asked
to record this when changing aligners), they were sent an automatic e-feedback and given background informa-
tion to increase the wearing time with the next aligner as “important information”.
All patients underwent standardized clinical and laboratory processes, as described in detail by Timm etal.14.
At the rst visit, a complete clinical examination, a full set of digital photographs and an intraoral scan as well
as radiographs according to the recommendations of the British Orthodontic Society were performed16. A basic
periodontal examination17,18 and a CMD screening19 were carried out in order to rule out contraindications to
CAT. Eventually, patients received their set of aligners required for CAT. e aligners were trimmed 2mm above
the gingival margin and the 1–1–2 treatment protocol, consisting of 3 consequent aligners per step (0.5mm for
one week, 0.625mm for one week, then 0.75mm for 2weeks), was followed. At this visit, patients were informed
about the importance of wearing the aligners for 22h per day and were instructed to check-in every aligner
change using the mobile app (including, as described, recording of the wearing time).
Figure1. Flowchart of the included groups.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
Vol.:(0123456789)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
A set of standardized intra- and extra-oral photographs were uploaded to the mobile application by patients
at least every two months for a follow-up to ensure close monitoring of treatment and to encourage patient
engagement.
Outcome and outcome measures. e outcome in this study was self-reported patient compliance,
recorded using the (application-based) questionnaire, measuring the daily aligner wearing time. Patients with
consistent use of the mobile application for aligner check-in and an aligner wearing time of 22h on 75% of their
aligners were classied as fully compliant, as dened previously14. Patients with inconsistent application usage
were classied as fairly or poorly compliant based on the aligner wearing time: Patients with the aligner wearing
time of 22h on 50–74.9% of their aligners were classied as fairly compliant and patients with an aligner wearing
time of 22h on only < 50% of their aligners as poorly compliant14.
ITS Analysis. A semi-monthly cross-section of recorded patient poor compliance was employed for our ITS
analysis. e data was partitioned by time, gender and age range (18 to 35, 36 to 55, and 56 to 64), resulting in 58
data points for the last six months of 2019 (pre-intervention cohort) and 64 data points for the last six months
of 2012 (post-intervention cohort), as recommended for ITS20. Because it was found that the patient compliance
was associated with a temporary slope change followed by a level change21 aer the introduction of the interven-
tion, the ITS was partitioned into three segments to account for this pattern: pre-intervention, post-intervention
segment A (2020-07-01 to 2020-09-15) and post-intervention segment B (2020-10-01 to 2020-12-15). Since the
outcome (poor compliance) is a count proportion, logistic regression was used for the analysis. e global model
was as follows:
where
• Yt is the proportion of poor compliance among patients who were treated in each halfmonth t, Xt is the vector
of the 8 covariates at time t, and E[Yt | Xt] is the expected value of Yt given the covariates Xt;
• timet is a continuous variable that represents the time elapsed (in halfmonths) from the start of the observa-
tion period;
• segment_At is an indicator variable for whether halfmonth t was in post-intervention segment A (1 if in
segment A, 0 otherwise);
• segment_A_timet is a continuous variable giving the time elapsed (in halfmonths) within post-intervention
segment A;
• segment_Bt is an indicator variable for whether halfmonth t was in post-intervention segment B (1 if in seg-
ment B, 0 otherwise);
• segment_B_timet is a continuous variable giving the time elapsed (in halfmonths) within post-intervention
segment B;
• gender_m is an indicator for gender (0 if female, 1 if male);
• age_36_55 is an indicator for age (1 if the age range is 36 to 55, 0 otherwise);
• age_56_64 is an indicator for age (1 if the age range is 56 to 64, 0 otherwise);
• β0 is an estimate of the log of the baseline odds; and
• each βi is an estimate of the log of the odds ratio of the respective covariate.
Some additional explanation of the last two points is given: e odds of an event with probability p is the
ratio p / (1 – p). In this case p is the probability of a patient being poorly compliant. If all the covariates are 0,
then the estimated odds of poor compliance are exp(β0). e other coecients relate to odds ratios. For instance,
exp(β6) is an estimate of the ratio of the odds given gender_m = 1 to the odds given gender_m = 0 (with all other
covariates equal), i.e. exp(β6) is a measure of how more/less likely male patients are to be poorly compliant than
female patients.
e methodology proposed by Burnham & Anderson22 was used, in particular multimodel inference (MMI).
e essential idea is to average over dierent models, rather than attempting to select a single best model. e key
ingredient is the Akaike information criterion (AIC), a measure of relative model quality founded on information
theory. To this end, all submodels of the global model were tted to the data; that is, for each of the 256 subsets of
{βi : 1 ≤ i ≤ n}, a model using only the selected covariates was tted to the data. (For brevity we conate covariates
with their coecients in the model.) e AIC of each model was calculated. Note that the small-sample version
of the AIC (AICc) was not used: Although the data seemingly consists of only 122 (= 58 + 64) points, each of
these points represents a group of patients. Indeed, the number of patients was used to weight the points in the
model tting, which is in fact equivalent to tting a logistic regression model on the level of individual patients,
with a response of either 1 (poorly compliant) or 0 (not poorly compliant). For each model, the dierence ∆ in
its AIC from the smallest AIC was calculated (so the top model has ∆ = 0). ese dierences were exponentiated
and then normalised to give the Akaike weight of each model. A subset of models was chosen on the basis of
the Akaike weights: e models {β0, β2, β3, β4, β6} and {β0, β2, β3, β4, β6, β8} had Akaike weights of 0.196 and
0.168 respectively, while the next highest weight was 0.099. ese two models were thus selected and their Akaike
weights renormalized. (Other selection criteria were explored, such as ∆ < 2 or the smallest subset such that the
sum of the Akaike weights was above 0.95, but each resulting averaged model suered from multicollinearity
E
[
Y
t|
X
t]=
logit
−
1
(β0+β1
x time
t+β2×
segment_A
t+β3×
segment_A_time
t+β4×
segment_Bt
+β
5
×
segment_B_timet
+β
6
×
gender_m
+β
7
×
age_36_55
+β
8
×
age_56_64
)
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
Vol:.(1234567890)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
amongst the covariates and in particular a non-positive-denite estimated covariance matrix.) e resulting
model, based on weighted averages over the selected models, is henceforth referred to as the MMI model.
e descriptive data analysis and pairwise testing were performed using JASP 0.14.1 (University of Amster-
dam, Amsterdam, e Netherlands). e ITS analysis was performed using version 0.13.1 of the Python library
statsmodels24. e pre-and post-intervention groups were pairwise compared with a two-sided Chi-square test
or a t-test for independent samples. p values less than 0.05 were considered statistically signicant.
Results
Cohort characteristics. Data of 5486 patients were available for analysis; of these, 1615 (29.4%) were male
and 3871 (70.6%) females. e median age at treatment initiation was 28 (range 18–64) years, the mean age
was 29.1 (range 18–64) years. e largest age group were young adults (aged 18–35years, n = 4456), followed
by adults (aged 36–55years, n = 984) and older adults (aged > 55years, n = 46). While the mean age was not
signicantly dierent between pre-and post-intervention periods, there was a signicant gender and age group
distribution dierence (p < 0.05) between the cohorts (Table1).
Compliance. Patients were classied according to the compliance criteria into full, fair, and poor compli-
ance. Pre-intervention, 703/1899 (37.0%) patients showed full compliance, 729/1899 (38.4%) fair compliance,
and 467/1899 (24.6%) poor compliance, whereas post-intervention 2382/5486 (43.4%) of patients showed full
compliance, 2262/5486 (41.2%) fair compliance, and 842/5486 (15.3%) poor compliance. Compliance was
higher in males than females (p < 0.001). No signicant dierence in compliance was found between age groups
(Table2).
Given that gender, but not age group distribution was signicantly associated with compliance, we further
compared compliance in both cohorts stratied by gender (Table3). We conrmed that compliance was signi-
cantly higher in the post-intervention than the pre-intervention group in both genders.
Descriptive plots of poor compliance over time are given in Figs.2, 3 and 4. e coecients of the MMI model
are shown in Table4. Note that p-values are not given, since doing so would conate dierent analysis paradigms;
instead, the Akaike weights of the coecients can be used as a measure of the relative strength of evidence. Fig-
ure5 shows the MMI model by starting date of treatment. e values were calculated by taking the mean of each
covariate at the given starting date and then using the coecients in Table4 to make a point estimate. e 95%
condence band was calculated using the estimated covariance matrix of the MMI model (not shown). Similarly,
the gures given in the Abstract—24.47% (95% CI: 22.59% to 26.46%) poor compliance in the pre-intervention
period and 9.32% (95% CI: 8.31% to 10.45%) in post-intervention period segment B—were calculated by taking
the mean of each covariate within each segment and using these as inputs for the MMI model. e results of the
MMI model suggest that the introduction of e-reminders had the desired eect, namely reducing poor compli-
ance. Note in Table4 that the Akaike weight for age_55_64 is lower than the weights of the other coecients.
Table 1. Cohort characteristics, stratied into pre-and post-intervention periods.
Covariates Pre-intervention Post-intervention p value
No. of subjects (n) 1899 5486
Age (mean ± SD) 28.8 ± 7.4 29.1 ± 8.1 0.13
Gender, female / male (%) 75.6% / 24.4% 70.6% / 29.4% < 0.001
Age 18- to 35- years old (%)
Age 36- to 55- years old (%)
Age 56- to 64- years old (%)
1601 (84.3%)
286 (15.1%)
12 (0.6%)
4456 (81.2%)
984 (17.9%)
46 (0.8%) < 0.05
Table 2. Compliance in the pre-and post-intervention cohorts, stratied by age group and by gender. A Chi-
square test was used to test for dierences in the strata in each cohort.
Pre-Intervention Post-Intervention
Overall
sample Full
compliance Fair
compliance Poor
compliance Chi-square Overall
sample Full
compliance Fair
compliance Poor
compliance Chi-square
Male 463 (24.4%) 185 (9.7%) 197 (10.4%) 81 (4.3%) X2 (2,
n = 1899) = 16.72
p = 0.000233
(p = < 0.001)
1615 (29.4%) 763 (13.9%) 648 (11.8%) 204 (3.7%) X2 (2,
n = 5486) = 19.39
p = 0.000061
(p = < 0.001)
Female 1436 (75.6%) 518 (27.3%) 532 (28.0%) 386 (20.3%) 3871 (70.6%) 1619 (29.8%) 1614 (29.7%) 638 (11.7%)
18- to 35-
years old 1601 (84.3%) 588 (31.0%) 616 (32.4%) 397 (20.9%)
X2 (4,
n = 1899) = 2.80
p = 0.591
4456 (81.2%) 1925 (35.1%) 1843 (33.6%) 688 (12.5%)
X2 (4,
n = 5486) = 1.28
p = 0.864
36- to 55-
years old 286 (15.1%) 112 (5.9%) 109 (5.7%) 65 (3.4%) 984 (17.9%) 438 (8.0%) 401 (7.3%) 145 (2.6%)
56- to 64-
years old 12 (0.6%) 3 (0.2%) 4 (0.1%) 5 (0.3%) 46 (0.8%) 19 (0.3%) 18 (0.3%) 9 (0.2%)
Tot al 1899 (100%) 703 (37.0%) 729 (38.4%) 467 (24.6%) 5486 (100%) 2382 (43.4%) 2262 (41.2%) 842 (15.3%)
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
Vol.:(0123456789)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
Moreover, the sign of coecient is not stable over the 95% condence interval. Both of these points are consistent
with the ndings of the chi-squared test that age was not signicantly associated with compliance.
Over the pre-intervention period, no signicant dierence or trend in compliance was identied (Fig.5). In
comparison, aer the introduction of the e-reminders and e-feedback (i.e., post-intervention), the percentage
of poorly compliant patients decreased substantially and remained consistently and signicantly lower than
pre-intervention. e downward trend did not continue indenitely; Fig.5 shows that the trend levelled o in
October 2020, i.e., a temporary slope was followed by a level change as described. Note that the initial increase
in post-intervention segment A is consistent with a seasonal eect since both July 2019 and July 2020 saw peaks
in poor compliance.
Discussion
is is the rst ITS study using real-world clinical data to investigate the impact of using e-reminders and
e-feedback via a mobile application during CAT. Our results showed that the introduction of e-reminders and
e-feedback signicantly and sustainably improved compliance. While compliance was generally higher in males
than females, the positive eect of the intervention was found in both males and females alike. Orthodontists and
dentists may want to employ mobile reporting options (like apps) to identify low compliance in CAT patients
and to address this low compliance using e-reminders.
Although a randomized controlled trial is considered the gold standard in attempting to prove causality and
to demonstrate the eectiveness of a new intervention, such trials are not always feasible nor needed25,26. For
instance, the running costs and time required for conducting trials, the associated bias by selection, and attrition
are among the limitations of randomized controlled studies26–28. In our study, we instead opted for an ITS design,
which is one of the strongest quasi-experimental designs26 not requiring randomization and having been found to
Table 3. Compliance within gender groups pre-and post-intervention. A Chi-square test was used to test for
dierences between intervention groups.
Full Compliance Fair Compliance Poor Compliance Chi-Square
Female
Pre-intervention 518 (36.1%) 532 (37.0%) 386 (26.9%) X2 (2, n = 5307) = 72.89
p = < 0.0001
Post-intervention 1619 (41.8%) 1614 (41.7%) 638 (16.5%)
Male
Pre-intervention 185 (40.0%) 197 (42.5%) 81 (17.5%) X2 (2, n = 2078) = 10.91
p = 0.0042 (< 0.05)
Post-intervention 763 (47.2%) 648 (40.1%) 204 (12.6%)
Figure2. Mean poor compliance by starting date of treatment. Note that this gure contains data from the
mid-period (January to June 2020), which was excluded from the statistical analysis.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
Vol:.(1234567890)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
have high internal validity20,21,29,30. Using the ITS approach on routine data allowed to demonstrate the eects of
the e-intervention on compliance in a large real-life population, i.e., with high generalizability and applicability.
In this study among CAT patients, we analysed factors inuencing patient compliance and trends before and
aer an intervention with e-reminders that emphasized the importance of compliance in wearing time. Since
the intervention (a new version of the mobile application, now including e-reminders, etc.) was deployed during
Figure3. Mean poor compliance by starting date of treatment, split by gender. e dashed vertical line
indicates the gap between the two segments.
Figure4. Mean poor compliance by starting date of treatment, split by age range. e dashed vertical line
indicates the gap between the two segments. e age range 56–64 was excluded from the gure due to the low
number of patients in this age group.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
Vol.:(0123456789)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
the rst quarter of 2020, the period between January 2020 and June 2020 included both patients who had started
their treatment with the old app version and those starting with the new one; this period was therefore excluded
to ensure consistency. Based on that and to control for seasonality, the period from January 2019 to June 2020
was excluded as well.
Although the gender and age group distribution diered signicantly between the pre-intervention and post-
intervention groups, we did not nd this to impact on compliance: Age was not signicantly associated with
compliance, and the combined eect of gender and intervention on compliance was analysed accordingly using
stratication analysis. We found the proportion of individuals with poor compliance to decrease signicantly in
both males and females aer the introduction of the e-reminders and e-feedback.
To our knowledge, there is no published ITS study investigating the eect of an intervention on patient com-
pliance during CAT, nonetheless, the ndings can be compared to a range of other reports. Reminders have been
found to have a positive eect on compliance during orthodontic treatment based on the results of two systematic
reviews, which is in line with our study8,31. Our ndings are also in agreement with a randomized trial showing
signicantly increased levels of oral hygiene compliance in orthodontic patients that received verbal education
on their treatment during their rst visits7. Added to that, the positive eect of active reminders established in
our study is also similar to the nding by Eppright etal., where weekly text message reminders highlighting the
importance of oral hygiene were shown to improve oral hygiene compliance in orthodontic patients6. Moreover,
the eect of regular positive reinforcement on general oral health through text messaging established by Jadhav
etal. is also in agreement with our ndings13.
Among many methods, reminding the patient on a regular basis to adhere to the treatment protocol has
been demonstrated to increase compliance before32,33. Patients found the mobile applications easy to use and
Table 4. e coecients of the MMI model.
Coecient 95% condence interval Akaike weight
Constant term (β0)− 1.03 (− 1.14, − 0.92) 1.00
Segment A (β2) 0.31 (0.11, 0.50) 1.00
Segment A time (β3)− 0.35 (− 0.46, − 0.24) 1.00
Segment B (β4)− 1.13 (− 1.29, − 0.97) 1.00
Gender_m (β6)− 0.40 (− 0.55, − 0.26) 1.00
Age_56_64 (β8) 0.42 (− 0.19, 1.04) 0.46
Figure5. e MMI model by starting date of treatment. e purple line gives the point estimates of expected
poor compliance while the purple shaded area is the 95% condence band. Each point indicates the mean poor
compliance of the given group, whereby the size of the point is proportional to the number of patients contained
within that group. Points with fewer than 5 patients were excluded from the gure. e dashed vertical line
indicates the gap between the two segments.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

8
Vol:.(1234567890)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
demonstrated the ability to use the app34,35, preferred the smartphone app intervention to other devices35, and
were satised with in-app functions that promote patient compliance with e-reminders36. e tendency to over-
estimate one’s own wearing times is reduced once patients know that the wearing time is monitored37. In addi-
tion to sending e-feedback and e-reminders to patients on regular basis, nancial incentives were reported to
enhance patient compliance38. In the context of the mobile application, a gaming component could be considered
to further optimize compliance39,40. e ecacy of these methods in CAT patients needs to be further studied.
is study has a number of strengths and limitations. First, it is one of the rst reports to demonstrate the
eect of e-reminders on improving patient compliance during CAT treatment. is nding is particularly signi-
cant as the literature indicates that orthodontic patients seeking treatment with removable appliances oen fail
to adhere to wearing times41–43 or sometimes do not complete the prescribed therapy at all44. Another strength
is that it used real-world data and employed the ITS approach. By doing so, high generalizability and applicabil-
ity, as well as a relatively large sample size, were combined with high internal validity. Among the limitations of
this study is the exclusion of the January 2020 to June 2020 group due to the impracticability of deploying the
intervention for all the patients at the same time. However, as with many interventions, a transition period is
needed for the intervention to produce the anticipated eect. is lag period could be excluded from the analysis
to avoid faulty interpretation of the analysis20. e ITS analysis also had limitations: e decision to partition the
data into three segments was post-hoc, although it was necessary to correctly model the data since a two-segment
approach cannot reect a temporary slope followed by a level change. Moreover, it would have been useful to have
a longer pre-intervention time period to build on more time points and capture possible patterns of seasonality
and dierentiate them from outliers. A longer pre-intervention period may also help to alleviate issues with
multicollinearity (see Sect.ITS Analysis). As a result, and to tackle the aspect of seasonality, we only included
individuals over similar time periods in 2019 (pre-intervention) and 2020 (post-intervention). We showed that
during summer, compliance decreased, possibly due to dierent cognitive distractions during periods of good
weather45. Another limitation is that the aligner wearing hours were reported by the patients using self-reports,
which are susceptible to biases. More in-depth data analyses and, possibly, qualitative investigations are needed
to address these limitations in detail.
Conclusion
Within the limitations of this study, e-reminder and e-feedback were eective measures for increasing compli-
ance in CAT patients. Orthodontists and dentists may consider digital monitoring and e-reminders to improve
compliance and thereby increase treatment success.
Data availability
e data is available upon request to the corresponding author.
Received: 7 January 2022; Accepted: 19 September 2022
References
1. Mehra, T., Nanda, R. & Sinha, P. K. Orthodontists’ assessment and management of patient compliance. Angle Orthod. 68, 155–222
(1998).
2. Richter, D. D., Nanda, R. S., Sinha, P. K., Smith, D. W. & Currier, G. F. Eect of behavior modication on patient compliance in
orthodontics. Angle Orthod. 68, 123–132 (1998).
3. Al-Abdallah, M., Hamdan, M. & Dar-Odeh, N. Traditional vs digital communication channels for improving compliance with
xed orthodontic treatment: A randomized controlled trial. Angle Orthod. 91, 227–235 (2021).
4. Siddiqui, N. R., Hodges, S. J. & Sharif, M. O. Orthodontic apps: an assessment of quality (using the Mobile App Rating Scale
(MARS)) and behaviour change techniques (BCTs). Prog. Orthod. 22(1), 25. https:// doi. org/ 10. 1186/ s40510- 021- 00373-5 (2021).
5. Huang, J., Yao, Y., Jiang, J. & Li, C. Eects of motivational methods on oral hygiene of orthodontic patients A systematic review
and meta-analysis. Med. (United States) 97, e13182 (2018).
6. Eppright, M., Shro, B., Best, A. M., Barcoma, E. & Lindauer, S. J. Inuence of active reminders on oral hygiene compliance in
orthodontic patients. Angle Orthod. 84, 208–213 (2014).
7. Cozzani, M. et al. Oral hygiene compliance in orthodontic patients: a randomized controlled study on the eects of a post-treatment
communication. Prog. Orthod. 17(1), 41. https:// doi. org/ 10. 1186/ s40510- 016- 0154-9 (2016).
8. Mohammed, H., Rizk, M. Z., Wafaie, K., Ulhaq, A. & Almuzian, M. Reminders improve oral hygiene and adherence to appoint-
ments in orthodontic patients: A systematic review and meta-analysis. Eur. J. Orthod. 41, 204–213 (2019).
9. Fenerty, S. D., West, C., Davis, S. A., Kaplan, S. G. & Feldman, S. R. e eect of reminder systems on patients’ adherence to treat-
ment. Patient Prefer. Adherence 6, 127–135 (2012).
10. Kannisto, K. A., Koivunen, M. H. & Välimäki, M. A. Use of mobile phone text message reminders in health care services: A nar-
rative literature review. J. Med. Internet Res. 16, e222 (2014).
11. Schwebel, F. J. & Larimer, M. E. Using text message reminders in health care services: A narrative literature review. Internet Interv.
13, 82–104 (2018).
12. Fjeldsoe, B. S., Marshall, A. L. & Miller, Y. D. Behavior change interventions delivered by mobile telephone short-message service.
Am. J. Prev. Med. 36, 165–173 (2009).
13. Jadhav, H. C. et al. Eect of Reinforcement of oral health education message through short messaging service in mobile phones:
A quasi-experimental trial. Int. J. Telemed. Appl. 2016, 7293516. https:// doi. org/ 10. 1155/ 2016/ 72935 16 (2016).
14. Timm, L. H., Farrag, G., Baxmann, M. & Schwendicke, F. Factors inuencing patient compliance during clear aligner therapy: A
retrospective cohort study. J. Clin. Med. 10, 3103 (2021).
15. Vandenbroucke, J. P. et al. Strengthening the reporting of observational studies in epidemiology (STROBE): Explanation and
elaboration. Int. J. Surg. 12, 1500–1524 (2014).
16. Isaacson, K. G., om, A. R., Atack, N. E., Horner, K. & Whaites, E. Orthodontic Radiographs: Guidelines for the Use of Radiographs
in Clinical Orthodontics (British Orthodontic Society, London, 2015).
17. Corbet, E. F. Oral diagnosis and treatment planning: Part 3. Periodontal disease and assessment of risk. Br. Dent. J. 213, 111–121
(2012).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

9
Vol.:(0123456789)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
18. Dietrich, T. et al. Periodontal diagnosis in the context of the 2017 classication system of periodontal diseases and conditions—
implementation in clinical practice. Br. Dent. J. 226, 16–22 (2019).
19. Ahlers, M. O. & Jakstat, H. A. Evidence-based development of a diagnosis-dependent therapy planning system and its implementa-
tion in modern diagnostic soware. Int. J. Comput. Dent. 8, 203–219 (2005).
20. Wagner, A. K., Soumerai, S. B., Zhang, F. & Ross-Degnan, D. Segmented regression analysis of interrupted time series studies in
medication use research. J. Clin. Pharm. er. 27, 299–309 (2002).
21. Bernal, J. L., Cummins, S. & Gasparrini, A. Interrupted time series regression for the evaluation of public health interventions: A
tutorial. Int. J. Epidemiol. 46, 348–355 (2017).
22. Burnham, K. P. & Anderson, D. R. Multimodel inference: Understanding AIC and BIC in model selection. Sociol. Methods Res.
33, 261–304 (2004).
23. Vanburen, J., Cavanaugh, J., Marshall, T. & Levy, S. AIC identies optimal representation of longitudinal dietary variables. J. Public
Health Dent. 77(4), 360–371 (2017).
24. Seabold, S. & Perktold, J. Statsmodels: Econometric and Statistical Modeling with Python. Proceedings of the 9th Python in Science
Conference 92–96 (2010). https:// doi. org/ 10. 25080/ majora- 92bf1 922- 011.
25. Hariton, E. & Locascio, J. J. Randomised controlled trials—the gold standard for eectiveness. BJOG An Int. J. Obstet. Gynaecol.
125, 1–4 (2018).
26. Hudson, J., Fielding, S. & Ramsay, C. R. Methodology and reporting characteristics of studies using interrupted time series design
in healthcare. BMC Med. Res. Methodol. 19, 1–7 (2019).
27. He, Z. et al. Clinical trial generalizability assessment in the big data era: a review. Clin. Transl. Sci. 13, 675–684 (2020).
28. Krauss, A. Why all randomised controlled trials produce biased results. Ann. Med. 50, 312–322 (2018).
29. Kontopantelis, E., Doran, T., Springate, D. A., Buchan, I. & Reeves, D. Regression based quasi-experimental approach when ran-
domisation is not an option: Interrupted time series analysis. BMJ 350, 1–4 (2015).
30. Shadish, W., Cook, T. & Campbell, T. Experiments and generalized causal inference. Exp. Quasi-Exp. Des. Gen. Causal Inference
100, 1–81 (2005).
31. Lima, I. F. P. et al. Inuence of reminder therapy for controlling bacterial plaque in patients undergoing orthodontic treatment: A
systematic review and meta-analysis. Angle Orthod. 88, 483–493 (2018).
32. Hardy, H. et al. Randomized controlled trial of a personalized cellular phone reminder system to enhance adherence to antiret-
roviral therapy. AIDS Patient Care STDS 25, 153–161 (2011).
33. Strandbygaard, U., omsen, S. F. & Backer, V. A daily SMS reminder increases adherence to asthma treatment: A three-month
follow-up study. Respir. Med. 104, 166–171 (2010).
34. Huang, Z. et al. A smartphone app to improve medication adherence in patients with type 2 diabetes in Asia: Feasibility randomized
controlled trial. JMIR mHealth uHealth 7, e14914 (2019).
35. Goldstein, C. M. et al. Randomized controlled feasibility trial of two telemedicine medication reminder systems for older adults
with heart failure. J. Telemed. Telecare 20, 293–299 (2014).
36. Grindrod, K. A., Li, M. & Gates, A. Evaluating user perceptions of mobile medication management applications with older adults:
A usability study. JMIR mHealth uHealth 2, e11 (2014).
37. Pauls, A., Nienkemper, M., Panayotidis, A., Wilmes, B. & Drescher, D. Eects of wear time recording on the patient’s compliance.
Angle Orthod. 83, 1002–1008 (2013).
38. Giurida, A. & Torgerson, D. J. To Enhance patient compliance. Heal. San Fr. 315, 703–707 (1997).
39. Greenstein, J., Topp, R., Etnoyer-Slaski, J., Staelgraeve, M. & McNulty, J. Eect of a mobile health app on adherence to physical
health treatment: retrospective analysis. JMIR Rehabil. Assist. Technol. 8, e31213 (2021).
40. De Simoni, A. et al. Electronic reminders and rewards to improve adherence to inhaled asthma treatment in adolescents: A non-
randomised feasibility study in tertiary care. BMJ Open 11, 1–11 (2021).
41. Tsomos, G., Ludwig, B., Grossen, J., Pazera, P. & Gkantidis, N. Objective assessment of patient compliance with removable ortho-
dontic appliances: A cross-sectional cohort study. Angle Orthod. 84, 56–61 (2014).
42. Al-Moghrabi, D., Salazar, F. C., Pandis, N. & Fleming, P. S. Compliance with removable orthodontic appliances and adjuncts: A
systematic review and meta-analysis. Am. J. Orthod. Dentofac. Orthop. 152, 17–32 (2017).
43. Arponen, H., Hirvensalo, R., Lindgren, V. & Kiukkonen, A. Treatment compliance of adolescent orthodontic patients with headgear
activator and twin-block appliance assessed prospectively using microelectronic wear-time documentation. Eur. J. Orthod. 42,
180–186 (2020).
44. Martin, C. A., Dieringer, B. M. & McNeil, D. W. Orthodontic treatment completion and discontinuation in a rural sample from
North Central Appalachia in the USA. Front. Public Heal. 5, 1–6 (2017).
45. Lee, J. J., Gino, F. & Staats, B. R. Rainmakers: Why bad weather means good productivity. J. Appl. Psychol. 99, 504–513 (2014).
Author contributions
L.H.T.: Conceptualization, Methodology, Formal analysis, Writing -Original Dra, Writing—Review and Edit-
ing. G.F.: Conceptualization, Writing -Original Dra, Writing—Review and Editing. D.W.: Formal analysis,
Writing—Review and Editing. M.B.: Writing—Review and Editing. F.S.: Conceptualization, Writing—Review
and Editing, Supervision.
Competing interests
Lan Huong Timm and Gasser Farrag declare current gainful employment by Sunshine Smile GmbH, the brand
owner of PlusDental. Daniel Wolf declares previous gainful employment by Sunshine Smile GmbH. Martin
Baxmann and Falk Schwendicke are members of the Scientic Board of the Sunshine Smile GmbH.
Additional information
Correspondence and requests for materials should be addressed to L.H.T.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

10
Vol:.(1234567890)
Scientic Reports | (2022) 12:16652 | https://doi.org/10.1038/s41598-022-20820-5
www.nature.com/scientificreports/
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2022
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com