ORIGINAL CONTRIBUTIONS

Mobile health application for remote

oral cancer surveillance
Praveen N. Birur, MDS; Sumsum P. Sunny, MDS;
Sidhant Jena, MS; Uma Kandasarma, MDS;
Shubhasini Raghavan, MDS; Bhanushree Ramaswamy, MDS;
Sunitha Perumal Shanmugam, BDS; Sanjana Patrick, BDS;
Rohan Kuriakose; Janhavi Mallaiah, MBBS;
Amritha Suresh, PhD; Radhika Chigurupati, DMD, MS;
Rani Desai, MA; Moni Abraham Kuriakose, MD

ighty percent of the worldwide burden of oral

cancer occurs in low-resource settings with
poor access to oral cancer specialists.1,2 This
is one of the primary reasons for delayed
diagnosis and poor outcome. The stage of disease at
diagnosis is the single most important factor that determines the treatment outcome of oral cancer patients.
The estimated age-adjusted incidence rate of oral cancer
in a low-resource setting such as India is 20 of 100,000,
which is one of the highest in the world.3 Oral cancer
accounts for 23% of all cancer-related deaths in the
Indian subcontinent.4 A system that offers specialist
consultation for dentists and health workers in lowresource settings may help identify high-risk patients
susceptible to oral cancer or to diagnose the cancer at
an early stage.
The use of wireless networking and telemedicine
has been investigated for improving disease surveillance
and to provide remote specialist consultation. The
addition of Web-based evaluation has improved breast
cancer screening by mammography and detection of
disease relapse in lung cancer patients.5,6 The extensive
network of mobile phones prevalent across populations
of all socioeconomic strata makes it a logical mode to
develop a similar system for screening and early detection of oral cancer. It has been previously demonstrated
that mouth self-examination7 or examination by trained
personnel8 can down-stage oral cancer and signicantly
improve survival rates in a high-risk population. An

ABSTRACT
Background. To determine the effectiveness of a mobile
phonebased remote oral cancer surveillance program
(Oncogrid) connecting primary care dental practitioners
and frontline health care workers (FHW) with oral cancer
specialists.
Methods. The study population (N 3,440) included a
targeted cohort (n 2,000) and an opportunistic cohort
(n 1,440) screened by FHW and dental professionals,
respectively. The authors compared the screening efcacy
in both groups, with specialist diagnosis considered the
reference standard. The outcomes measured were lesion
detection and capture of interpretable images of the oral
cavity.
Results. In the targeted cohort, among 51 of 81 (61%)
interpretable images, 23 of 51 (45%) of the lesions were
conrmed by specialists, while the opportunistic cohort
showed 100% concordance with the specialists (106 of 106).
Sixty-two of 129 (48%) of the recommended patients underwent biopsy; 1 of 23 (4%) were in the targeted cohort,
and 61 of 106 (57%) were in the opportunistic cohort.
Ninety percent of the lesions were conrmed to be malignant or potentially malignant.
Conclusions. The mobile healthbased approach adopted in this study aided remote early detection of oral cancer
by primary care dental practitioners in a resourceconstrained setting. Further optimization of this program
is required to adopt the system for FHW. Evaluation of its
efcacy in a larger population is also warranted.
Practical Implications. The increased efciency of
early detection by dentists, when assisted by a remote
mobile healthbased approach, is a step toward a more
effective oral cancer screening program.
Key Words. Oral cancer; early detection; screening;
mobile health; health workers; remote assistance; primary
care dentist.
JADA 2015:-(-):--http://dx.doi.org/10.1016/j.adaj.2015.05.020

Copyright 2015 American Dental Association. All rights reserved.

JADA

( )

- -

http://jada.ada.org

FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

2015 1

ORIGINAL CONTRIBUTIONS

Frontline
health care
workers

Frontline
health care
workers

Frontline
health care
workers

Nodal center
Primary
health center

Targeted Cohort

(Frontline Health Care Workers)

Opportunistic Cohort
(Dental Surgeons)

Co-coordinating
cancer center

Frontline
health care
workers

print & web 4C=FPO

Frontline
health care
workers

Nodal center

Nodal center

Primary
health center

Dental college
or clinic

Tertiary cancer center

Remote consultation with an oral
cancer specialist
Dentist
Tertiary cancer center
Histologic evaluation of
confirmed patients

Frontline
health care
workers

Figure 1. Node-and-hub model and overall study design. A. Study was carried out with a central coordinating center and multiple nodal centers.
B. Study design included 2 patient cohorts, targeted and opportunistic; images from both cohorts were sent to the tertiary care center for remote
consultation and histologic evaluation.

improvisation of this program by providing remote

specialist consultation is likely to improve scalability
of this concept. In this project, we attempted to develop a
mobile phonebased platform for risk stratication and
evaluation of the mouth, which is remotely monitored
by oral cancer specialists. We hope that adapting this
technology in a low-resource, high-risk population will
be a step toward down-staging oral cancer.
METHODS

Study design. Oncogrid is a mobile phonebased cancer

surveillance program developed as a hub-and-node
based model (Figures 1 and 2). The hub is located at a
tertiary care cancer center (Mazumdar-Shaw Cancer
Center, Bangalore, India), and the nodes are located in 2
different settings: dentists in a dental clinic and frontline
health care workers (FHW) associated with a primary
health center. The project was approved by the relevant
institutional research and ethics committees. Informed
consent was obtained from all participating patients. The
work ow involved the use of an Android mobile phone
(HTC Wildre S, 5MP camera) and oral cancer screening
software equipped with a clinical decision algorithm for

2 JADA

( )

- -

http://jada.ada.org

risk assessment and diagnosis (Figure 1). This study was

conducted over 12 months (January to December 2010).
Study population and screening. Cohort 1, the targeted group, consisted of participants from 2 rural villages, Anakanur (n 2,055) and Poshetahalli (n 3,184),
of the Chikkabalapur district in the suburbs of Bangalore
City. These participants, who had a high prevalence of
oral cancer risk habits, were selected on the basis of a
prior health survey carried out by a voluntary nongovernmental organization (Biocon Foundation, Bangalore,
India). The targeted cohort, consisting of all eligible
participants in this dened population, was screened by
the FHW (n 4) (associated with the primary health
center) in a door-to-door manner. Cohort 2 included
participants screened by dental surgeons in a primary
dental clinic. Because the screening was conducted in
patients attending the clinic (KLES Institute of Dental
Sciences, Bangalore) for other treatment, the cohort was
termed the opportunistic group. In both cohorts, the

ABBREVIATION KEY. FHW: Frontline health care worker.

GPRS: General packet radio service. mHealth: Mobile health.

2015
FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

triaging of the participants as members of a

Server
high-risk group was
Sana
based on the presence of
Open-source
mobile
any of the following 3
health platform
factors: history of risk
habits (smoking tobacco,
chewing betel leaf or
OpenMRS
Open-source medical
gutka [combination of
record program
areca nut with or without
tobacco], or regular use
of alcohol), older than 40
years, and clinical signs
(nonhealing ulcers, red
or white patches of the
mouth that last more
than 3 weeks, restriction
of mouth opening, and
swelling of the neck).
The dentists and
FHW involved in this
Oral cancer specialist at the
Frontline health care workers
project underwent a
cancer center
structured training program by oral cancer
specialists on risk factors
Figure 2. Oncogrid technology architecture. Overall pipeline of Oncogrid program is depicted; patient details and
and screening methods
images are captured through Sana. Data are then uploaded to the OpenMRS system and subsequently accessed
to identify suspicious le- by specialists at the tertiary care center. Feedback from the specialists is sent back to frontline health care workers.
sions using interactive
education materials. The health care providers were
store data for individual patients. The latter includes a
trained to examine different subsites of the oral cavity in concept dictionary, which can be used to dene and
a systematic fashion. These included right gingivobuccal gather multiple observations from a single encounter
mucosa and retromolar trigone, lower labial mucosa, left and also has built-in support modules, which is key in
gingivobuccal mucosa and retromolar trigone, upper
implementing Sana in rural areas.
labial mucosa, palate, oor of the mouth, and oral
At the front end, Sana, downloaded onto the phone, is
tongue. The tongue was further divided into dorsal,
used for media and data (audio, image, location-based
ventral, right, left, and lateral borders. The FHWs were data, and text) capture, and at the back end provides an
also trained to specically examine the region in the oral intuitive user interface for management of the media.
cavity where the gutka or tobacco was placed for chew- Sana also allows modularity and interoperation, allowing
ing. The specialists providing remote consultation were multiple front-end data to be accessed by the coordinaasked to report the lesions as nonneoplastic, potentially tion center. The software also includes built-in training
malignant, or malignant.
and work ow solutions at the mobile phone level. Data
The patients who were diagnosed with oral lesions in transfer is carried out using a number of interfaces,
the targeted cohort were monitored by the FHW once
including general packet radio service (GPRS), wi-, text
every 3 months for 2 years by clinical examination and
messages, and USB tethering, and was controlled algofor habit cessation counseling. Patients with oral lesions rithmically to ensure reliable upload in regions with poor
recommended for biopsy by remote experts review were mobile network coverage. To avoid midupload failure of
referred to oral cancer specialists in the dental college or data containing large image les, Sana uses packetizato the tertiary care cancer center.
tion, which allows upload of les in chunks so that large
Technology platform. The mobile health (mHealth)
les can be sent to the server even with smaller bandwidth.
system used for the Oncogrid project was developed using
The synchronization between the server that hosts
the Sana platform (Computer and Articial Intelligence
Sana and the OpenMRS was through GPRS. The project
Laboratory, Massachusetts Institute of Technology, http:// also involved customization of OpenMRS through the
sana.mit.edu) and OpenMRS (http://openmrs.org), an
development of an Oncogrid-specic dashboard to
open-source medical record system. Sana is designed as a monitor the work ow, individual patient evaluation, and
feedback. The individual images were visualized in detail
work ow interface and is integrated with OpenMRS, a
Java-based Web application that can be used to input and via zooming and contrast adjustment. As the OpenMRS

JADA

( )

- -

http://jada.ada.org

FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

2015 3

print & web 4C=FPO

ORIGINAL CONTRIBUTIONS

ORIGINAL CONTRIBUTIONS

BOX

Questionnaire to assess clinical details of

patients.
TARGETED SCREENING COHORT
1. Enter patient medical record identication number
2. Enter patient rst name
3. Enter patient last name
4. Enter patient birthdate
5. Select patient sex
6. Enter patient mobile number
7. Health center
8. Age (40): yes or no
9. Family history of cancer: yes or no
10. Education level: none, high school, or college
11. Employed: yes or no
12. Drinker: no, social, or heavy
13. Smoker: no, social, or heavy
14. Paan or chewing tobacco: no, social, or heavy
15. Visit dentist: yes or no
16. Does the patient have or had any symptoms of oral lesion, lump in neck, or difculty
opening mouth: yes or no
17. Add picture.
18. Save and upload to server.
OPPORTUNISTIC SCREENING COHORT
1. Enter patient medical record identication number
2. Enter patient rst name
3. Enter patient last name
4. Enter patient birthdate
5. Select patient sex
6. Enter patient mobile number
7. Health center
8. Age (40): yes or no
9. Family history of cancer: yes or no
10. Education level: none, high school, or college
11. Employed: yes or no
12. Drinker: no, social, or heavy
13. Smoker: no, social, or heavy
14. Paan or chewing tobacco: no, social, or heavy
15. Visit dentist: yes or no
16. Does the patient have or had any symptoms of difculty opening mouth, mouth ulcer,
white or red patch in mouth. swelling in neck, or growth in mouth: yes or no.
If yes for mouth ulcer or sore: less than 3 weeks or more than 3 weeks; one or multiple;
painful, loss of sensation, or no pain; how is surrounding tissue: everted or not everted.

If yes for white patch: duration of patch: less than 3 weeks or more than 3 weeks;
cannot be wiped off or can be wiped off; how many patches: single or multiple; how is
the surface of the patch: irregular or uniform; what is the color of the patch: white or
whitish red; how does the patch feel: painful or no pain.

If yes for swelling in neck; less than 3 weeks or more than 3 weeks; consistency of
swelling in neck: rm or soft; xation of swelling in neck: mobile or xed; skin over
swelling in neck: smooth, normal, or irregular; location of swelling in neck: lateral or
midline; size of swelling in neck: less than 2 centimeters, 2 to 4 cm, 4 to 6 cm, or more
than 6 cm.

If yes for growth: less than 3 weeks or more than 3 weeks; any changes in size of
growth: increased, no change, or decreased; how does the growth feel: painful, loss
of sensation, or no pain; how is the consistency of the growth: rm or soft; how is the
surface of growth: irregular overlying mucosa or regular overlying mucosa; condition
of teeth around growth: loose or normal.

17. Add picture

18. Save and upload to server

4 JADA

( )

- -

http://jada.ada.org

is health level 7 standard messaging system

compatible, the data could be exported to
the hospital medical record system in cases
in which the patients required a visit to the
specialist center. The data could also be
exported standard statistical software
packages for detailed analysis and reporting. The overall technology platform
deployed for Oncogrid is depicted in
Figure 2.
Work ow. The Sana software on mobile phone generates patient registration
number, date of registration, and demographic details. Each participant was
interrogated using a risk evaluation questionnaire incorporated in the application.
The details obtained consisted of demographics, risk factors, and symptoms (Box).
It included the following questions: Is your
age greater than 40? Do you have a family
history of cancer? Do you drink alcohol? Do
you smoke cigarettes or beedi (local cigarette)? Do you chew paan (betel leaf) with
tobacco (gutka or areca nut)? Do you have
any white or red patches in your mouth? Do
you have any ulcers in your mouth? Do
you have any swelling of neck or difculty
in opening the mouth? If the answer to any
of these questions is yes, then the patient
was considered to be at high risk and was
enrolled onto the study after providing
informed consent. Intraoral lesions, if present, were photographed. The data obtained
through the mobile phones, along with
photographs of the suspicious lesions, were
uploaded to the OpenMRS system through
a secure server.
The system then created a queue that
was reviewed at the coordinating center
by the oral cancer specialist. The specialist
reviewed the image and judged it as interpretable or not interpretable. The interpretable images were clinically stratied as
nonneoplastic, potentially malignant, or
malignant. For oral potentially malignant
and malignant lesions, it was recommended
that the patients undergo biopsy. Text
messages (also known as short messaging
services, or SMS) with follow-up instructions were sent to the respective FHW
and dentists. On receiving the message, the
FHW and dentists referred patients with
suspected lesions to the nodal centers (primary health clinic or dental college) for
biopsy. Reporting errors at all levels were
managed effectively with digital storage and

2015
FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

ORIGINAL CONTRIBUTIONS

transfer of data to improve FHW and dentists efciency

in screening. A continuous feedback of the results was
provided to the FHW and dentists through mobile
phones. A Web-based dashboard also facilitated the program coordinators to detect bottlenecks and rectify them.
Histologic evaluation and follow-up. The histopathologic reporting of the oral lesion was done as
per the World Health Organization classication; oral
epithelial precursor lesions were classied into 6 categories (squamous cell hyperplasia; mild, moderate,
and severe dysplasia; carcinoma-in-situ; invasive carcinoma).9 For treatment decisions, we adopted the binary
system developed by Kujan and colleagues.10 Accordingly, nondysplastic lesions were combined with mild
dysplasia as low-risk lesions, while moderate and severe
dysplasia along with carcinoma-in-situ were grouped as
high-risk lesions. The biopsy results were also uploaded
via mobile phone by the nodal center to the OpenMRS.

Patients with nonneoplastic lesions were discharged

after antitobacco counseling while those with low-risk
lesions were counseled for habit cessation and prescribed
topical application of water-soluble vitamin A palmitrate
(Aquasol-A, USV Limited [Mumbai]; 50,000 United
States Pharmacopeia units per milliliters) twice daily for
3 months.11,12 Patients with high-risk lesions and invasive
cancer were referred to the tertiary care cancer center for
surgical excision and adjuvant treatment as per established cancer treatment guidelines.
Compliance with biopsy and treatment at nodal and
cancer centers were monitored through OpenMRS.
Long-term surveillance after completion of treatment
was carried out in the community in partnership with
the nodal center. All patients were educated about the
harmful effects of tobacco and areca nut. Patients with
high-risk habits were kept under surveillance by the
FHW to monitor compliance to counseling and the

Study
population

Cohort I
Targeted screening by
frontline health care workers
(n = 2,000)

Cohort II
Opportunistic screening
by dental professionals
(n = 1,440)

Patients in the
high-risk group
with lesion
(n = 130)

Patients in the
high-risk group
with lesion
(n = 106)

Patients involved
in mobile phone
screening
(n = 83)

print & web 4C=FPO

Interpretable
images
(n = 51)

Lesions negative on
remove diagnosis
(n = 28)

Patients not
involved in mobile
phone screening
(n = 47)

Noninterpretable
images
(n = 32)

Lesions positive on
remove diagnosis
(n = 23)

Patients involved
in mobile phone
screening
(n = 106)

Interpretable
images
(n = 106)

Lesions negative on
remove diagnosis
(n = 0)

Patients not
involved in mobile
phone screening
(n = 0)

Noninterpretable
images
(n = 0)

Lesions positive on
remove diagnosis
(n = 106)

Figure 3. Patient distribution across 2 cohortscohort 1 (targeted; n 2,000) and cohort 2 (opportunistic; n 1,440)and number of high-risk
lesions identied. All patients screened in cohort 2 consented to mobile screening; in cohort 1, 47 patients were not involved in screening. All 106
images were interpretable and diagnosed positively in cohort 2, while in cohort 1, 38% (n 32) of the images were noninterpretable, and 45% (n 23)
of identied lesions were conrmed positive.

JADA

( )

- -

http://jada.ada.org

FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

2015 5

ORIGINAL CONTRIBUTIONS

specialist. All statistical

calculations were carried
out by GraphPad software
(GraphPad Software).

print & web 4C=FPO

RESULTS

Patient cohort. The

total study population consisted of 3,440 patients,
with 2,000 patients assigned
to cohort 1 (screened by
FHW) and 1,440 to cohort
2 (screened by dental professionals). Patient distribution is provided in
Figure 3.
Targeted screening by
FHW. Among the 2,000
patients screened by FHW
(n 4), the female-to-male
ratio was high (3:1); patients
were within the age range
of 20 to 85 years. Targeted
screening identied 130 patients (6.5%) with suspicious lesions. Thirty-six
percent of these patients
(n 47) did not provide
consent for being photographed and were excluded
from the study. The images
for 83 accrued patients were
captured and transmitted
to the coordinating center
for consultation (Figure 4).
Images for 32 patients (38%)
were judged to be noninterpretable by the oral
cancer specialist at the
center. Among the 51 interpretable images, the
remote specialist conrmed
the diagnosis of 23 patients
with either potentially malignant (n 22) or malignant lesions (n 1). The
FHW were hence able to
detect the patients with
Figure 4. Representative images taken from opportunistic and targeted cohorts and from OpenMRS. Images
taken by dentists of patients in the opportunistic cohort (A, B) and by frontline health care workers in the
positive lesions with a postargeted cohort (C, D). D. Uninterpretable image taken by a frontline health care worker. E. Snapshot of the
itive predictive value of
OpenMRS system. Patient details are masked to protect patient condentiality.
45%, with the specialist
progression of the lesions, if any, using the Oncogrid
consultation being the reference standard. These patients
platform.
were referred to the nodal center for biopsy.
Statistical analysis. Statistical analysis was carried
Opportunistic screening by dental professionals.
out to determine the concordance of the diagnosis
Among the 1,440 patients screened by the dental proby FHW and dentists with remote diagnosis by the
fessionals in cohort 2, the male-to-female ratio was 3:1;

6 JADA

( )

- -

http://jada.ada.org

2015
FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

ORIGINAL CONTRIBUTIONS

TABLE
patients were within the age
range of 18 to 73 years. One
Comparison of outcome between targeted and
hundred six patients were idenopportunistic screening groups.
tied to have oral potentially
malignant or malignant lesions.
OPPORTUNISTIC
EVENT
TARGETED
SCREENING
SCREENING
All patients consented to and
(N [ 1,440)
(N [ 2,000)
were accrued for the study. The
NO.
%
NO.
%
images of all patients uploaded by
No. of High-risk Patients With Lesion
130
6.5
106
7.36
the dental professionals were
83
63.85
106
100.00
found to be of diagnostic quality No. of Patients Consented to Capture Images
51
61.45
106
100.00
(Figure 4, Table). The oral cancer No. of Interpretable Images
Remote Diagnosis Conrmation of Potentially
23
45.10
106
100.00
specialists at the coordinating
Malignant or Malignant Lesion by Specialists
center conrmed all the lesions
No. of Patients Who Underwent Histologic
1
4.35
61
57.55
to be potentially malignant
Evaluation
(n 101) or malignant (n 5).
Normal
0
0.00
6
9.84
The concordance and positive
Severe dysplasia
0
0.00
10
16.39
predictive value of dental proModerate dysplasia
0
0.00
15
24.59
fessionals interpretation was
Mild dysplasia
0
0.00
24
39.34
100%.
Nondysplastic lesions
0
0.00
1
1.63
Histologic evaluation. The
Squamous cell carcinoma
1
100.0
5
8.20
patients from both cohort 1
(n 23) and cohort 2 (n 106)
with high-risk lesions were referred for histologic eval- reported in cervical and breast cancer screening
uation. Out of these 129 patients, 62 (cohort 1, n 1;
programs.8,13,14
cohort 2, n 61) underwent biopsy; compliance in
The use of mobile phone technology has the potencohort 1 was 4% (1 of 23) compared with 57% (61 of 106) tial to bridge disparity in terms of access to health
in cohort 2. Seventy-nine percent of the biopsy results
care by combining visual screening with remote conwere reported as lesions with dysplasia (49 of 62), while sultation. Mobile diagnostic systems in combination
9.6% were malignant lesions (n 6). Nine percent (6 of with e-communication and off-site expert diagnosis
62) of the biopsy ndings were reported as nonmalighave been attempted at remote locations for early denant (no evidence of malignant or potentially malignant tection of lung and cervical cancers as an effort toward
changes) and 2% (n 1) were nondysplastic leukoplakia providing specialist expertise to inaccessible, low(epithelial hyperplasia without dysplasia). Among
resource settings.6,15 Accessibility for clinical examinathe dysplastic lesions, 49% were with mild dysplasia
tion, presence of well-dened risk factors, and clinical
(n 24), while others were moderate (31%, n 15) and features make oral cancers well suited for mobile phone
severe (20%, n 10). The patient referred from cohort 1 based screening. In this study, we report potential
was diagnosed with malignancy (Table). All the malig- effectiveness of mHealth-aided oral cancer screening
nant lesions (n 6) in both the cohorts were stage I
in targeted (by FHW) and opportunistic (by dental
professionals) cohorts.
(T1, N0, M0) lesions.
Among the dysplastic patients in cohort 2, 38%
A comparative analysis of screening in both the co(n 19) of patients showed compliance for follow-up
horts, aided by remote diagnosis of oral cancer specialists,
indicated that opportunistic screening (dentists) worked
in the dental clinic. None of these patients developed
clinical evidence of disease progression during the
better in terms of lesion detection, image quality, and
2 years of follow-up. All 23 patients in cohort 1 during
diagnostic accuracy. The FHW were able to detect lesions
the 2 years of follow-up also did not develop malignant in 5% of the patients and transmit images with 61% actransformation.
curacy, compared with the 100% accuracy achieved by the
dental surgeons. In mHealth-based studies, the dependence of diagnosis on the quality of the photographs is
DISCUSSION
an issue. However, the high efcacy of image capture
The low survival rates of oral cancer patients in lowby dental professionals indicated that the technology is
resource settings are primarily dependent on the stage
robust. In contrast, 38% of the images in the targeted
at diagnosis. The late presentation of patients is attrib- cohort were not interpretable, suggesting that further
uted to the limited access to specialized clinical services.
training of the FHW or newer technology that overcomes
Population-based oral cancer screening programs by vi- human error needs to be developed. We are in the process
sual inspection in low-resource settings have shown to
of developing an automated oral photography capture
reduce mortality by 34%.8 Similar benets have been
device to overcome this limitation.

JADA

( )

- -

http://jada.ada.org

FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

2015 7

ORIGINAL CONTRIBUTIONS

One of the issues in the program was the difculty in

evaluating the specicity of diagnosis in both the cohorts.
The conrmation of a negative diagnosis by on-site
specialists was not feasible because the program was
conceived as an outreach screening program. Nevertheless, the fact that 45% of the lesions detected by the FHW
were conrmed by the specialists indicated that they may
be an invaluable resource for door-to-door screening in
low-resource settings. The accuracy of detection increased
with remote diagnosis; 54% of the patients avoided unnecessary referral and biopsy.
This study also provided evidence that advanced
training in terms of oral cancer detection and mHealth
systems can aid FHW in carrying out oral cancer
screening and surveillance. In a lung cancer program,
when patients themselves were trained appropriately,
early detection of disease relapse was observed.6 On-site
diagnosis by trained nurses also showed high concordance with off-site experts in a cervical cancer screening
program.15 Remote connectivity to specialist centers has
also been successfully implemented in rural populations,
including India, with a denite clinical advantage in
nonmalignant disorders.16-21 Nevertheless, these projects
lacked the ability to reach out to asymptomatic patients
at their workplace or home, which is desirable in cancer
screening programs. However, it is to be noted that
all the cancer patients diagnosed in this study were stage
T1 (< 2 centimeters) disease, further attesting to the
potential of mHealth for detection of both early stage
malignant or potentially malignant lesions.
In a cervical cancer screening study, telemedicine
enabled effective diagnosis, thereby reducing delays
in referrals and the need for travel from distant sites.15
In our study, aided by remote consultation, 54% of the
patients were diagnosed as negative for the lesions when
assessed by the specialist, ensuring that only patients
with conrmed clinical lesions were referred to the tertiary care center. mHealth consultation is thus also an
effective strategy to increase the efcacy of the FHW
in terms of detecting high-risk lesions and avoiding
unnecessary referrals.
Compliance was a primary issue in the study, especially in the targeted screening group. Only 4% from
cohort 1 and 57% from cohort 2 underwent the recommended biopsy. Similar results were observed in other
studies, in which the compliance to biopsy was only
10%.7 Multiple factors such as denial of their medical
condition, asymptomatic nature of high-risk oral lesions,
invasiveness of the procedure, need to travel to the nodal
centers, and poor socioeconomic background contribute
to this poor compliance. As observed in other oral cancer
screening trials, multiple rounds of screening may be
required to improve compliance.22 In addition, alternative noninvasive approaches such as the use of molecular
imaging techniques23 and salivary biomarkers might
serve to address this issue. Nevertheless, mHealth has

8 JADA

( )

- -

http://jada.ada.org

shown potential to improve compliance through automated communication directly to patients (text messages, Web-based methods, tailored education), as has
been observed in many other studies, including colorectal cancer24 and breast cancer.5 On the basis of the
success of this project, we have initiated a randomized
study comparing the cost-effectiveness of mHealth
and conventional oral cancer screening in a larger
population.
One of the major challenges in the management of
oral potentially malignant lesions is the histologic interpretation of oral dysplasia and developing a decisionmaking algorithm based on the grade of dysplasia.
Although Holmstrup and colleagues25 reported various
rates of malignant transformation in nondysplastic (2% to
11%) and mild dysplastic lesions (11% to 14%), no correlation was observed between the levels of dysplasia and
malignant transformation. However, several other studies
have demonstrated severity of dysplasia to be a signicant
predictor for malignant transformation.10,26,27 To lower
interobserver variability and to make therapeutic decisions, a binary system of dysplasia grading was proposed by Kujan and colleagues.10 In this study, we have
used this binary system for the management of oral
potentially malignant lesions. According to this
decision-making algorithm, nondysplastic and mild
dysplastic lesions were managed conservatively by habit
cessation and chemoprevention, while moderate and
severe dysplastic lesions were treated by chemoprevention and surgical intervention in addition to habit
cessation.
CONCLUSIONS

This pilot telemedicine program is novel in that it

found that mobile phonebased oral cancer screening
and surveillance were feasible in low-resource settings. It
facilitated optimal utilization of professional resources
and early detection of asymptomatic oral neoplastic
lesions. In the long term, this strategy is expected to
down-stage oral cancer and improve survival rates.
With wide penetration of the mobile phone network,
including in remote locations, this project will have a
major impact in the way oral cancer is managed in the
community. In the future, we foresee that medical and
dental institutions equipped with knowledge in epidemiology, prevention, and screening, when remotely
integrated to a local community cancer screening service and a regional tertiary cancer center, will bring
about a paradigm change in cancer care and outcomes
in low-resource settings. As an initial step in this direction, we are in the process of developing a network
of mHealth-based oral cancer outreach screening
programs. n
Dr. Birur is a professor and head, Department of Oral Medicine and
Radiology, KLES Institute of Dental Sciences, Bangalore, India.

2015
FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:39 pm ! ce

ORIGINAL CONTRIBUTIONS

Dr. Sunny is a senior house ofcer, Head and Neck Oncology, Mazumdar
Shaw Cancer Center, and a doctoral student, Integrated Head and Neck
Oncology Program, Mazumdar Shaw Center for Translational Research,
Mazumdar Shaw Medical Foundation, Mazumdar Shaw Medical Center,
Bangalore, India.
Mr. Jena is the chief executive ofcer, Jena Care, Koramangala, Bangalore,
India.
Dr. Kandasarma is a professor and head, Department of Oral and
Maxillofacial Pathology, KLES Institute of Dental Sciences, Bangalore, India.
Dr. Raghavan is a reader, Department of Oral Medicine and Radiology,
KLES Institute of Dental Sciences, Bangalore, India.
Dr. Ramaswamy is a lecturer, Department of Oral Medicine and Radiology, KLES Institute of Dental Sciences, Bangalore, India.
Dr. Shanmugam is a surgeon, Dental Section, Government Hospital,
Hosur, India.
Dr. Patrick is a post graduate student, Department of Oral Medicine and
Radiology, KLES Institute of Dental Sciences, Bangalore, India.
Mr. Rohan Kuriakose is an undergraduate student, School of Engineering,
University of Buffalo, Buffalo, NY.
Dr. Mallaiah is a program ofcer, Biocon Foundation, Bangalore, India.
Dr. Suresh is a principal investigator, Integrated Head and Neck Oncology
Program, Mazumdar Shaw Center for Translational Research, Mazumdar
Shaw Medical Foundation, Mazumdar Shaw Medical Center, Bangalore,
India, and a research technologist, Mazumdar Shaw Cancer Centre, Roswell
Park Collaborative Head and Neck Oncology Research Program, Roswell
Park Cancer Institute, Buffalo, NY.
Dr. Chigurupati is an associate professor, Department of Oral and
Maxillofacial Surgery, Boston University, Boston, MA.
Ms. Desai is the head, Biocon Foundation, Bangalore, India.
Dr. Moni Abraham Kuriakose is a director, Department of Surgical
Oncology, Mazumdar Shaw Cancer Center, Narayana Health, and a professor and vice chairman, Department of Head and Nech, Plastic and
Reconstructive Surgery, Roswell Park Cancer Institute, Buffalo, NY. Address
correspondence to Dr. Kuriakose at Department of Surgical Oncology,
Mazumdar Shaw Cancer Center, Narayana Health, 258/A, Bommasandra
Indl Area, Anekal Taluk, Bangalore 560099, India, e-mail makuriakose@me.
com.
Disclosure. None of the authors reported any disclosures.
This project was funded by Krishak Bharati Co-operative Limited,
Gautham Budh Nagar, UP, India. The authors acknowledge the support
of Mazumdar Shaw Medical Foundation, Narayana Health City, and KLES
Institute of Dental Sciences, Bangalore, India.
1. Mignogna MD, Fedele S, Lo Russo L. The World Cancer Report and
the burden of oral cancer. Eur J Cancer Prev. 2004;13(2):139-142.
2. Warnakulasuriya S. Global epidemiology of oral and oropharyngeal
cancer. Oral Oncol. 2009;45(4-5):309-316.
3. Coelho KR. Challenges of the oral cancer burden in India. J Cancer
Epidemiol. 2012;2012:701932.
4. Dikshit R, Gupta PC, Ramasundarahettige C, et al. Cancer mortality in
India: a nationally representative survey. Lancet. 2012;379(9828):1807-1816.
5. Atlas SJ, Grant RW, Lester WT, et al. A cluster-randomized trial of a
primary care informatics-based system for breast cancer screening. J Gen
Intern Med. 2011;26(2):154-161.
6. Denis F, Viger L, Charron A, et al. Detection of lung cancer relapse
using self-reported symptoms transmitted via an internet web-application:
pilot study of the sentinel follow-up. Support Care Cancer. 2014;22(6):
1467-1473.

7. Elango KJ, Anandkrishnan N, Suresh A, et al. Mouth self-examination

to improve oral cancer awareness and early detection in a high-risk population. Oral Oncol. 2011;47(7):620-624.
8. Sankaranarayanan R, Ramadas K, Thomas G, et al. Effect of screening
on oral cancer mortality in Kerala, India: a cluster-randomised controlled
trial. Lancet. 2005;365(9475):1927-1933.
9. Barnes L, Eveson JW, Reichart PA, Siransky D, eds. Pathology and
Genetics of Head and Neck Tumours. Lyon: IARC Press; 2005. Available at:
http://www.iarc.fr/en/publications/pdfs-online/pat-gen/bb9/BB9.pdf.
Accessed May 29, 2015.
10. Kujan O, Oliver RJ, Khattab A, et al. Evaluation of a new binary
system of grading oral epithelial dysplasia for prediction of malignant
transformation. Oral Oncol. 2006;42(10):987-993.
11. Epstein JB, Gorsky M. Topical application of vitamin A to oral leukoplakia: a clinical case series. Cancer. 1999;86(6):921-927.
12. Piattelli A, Fioroni M, Santinelli A, Rubini C. bcl-2 expression and
apoptotic bodies in 13-cis-retinoic acid (isotretinoin)-topically treated oral
leukoplakia: a pilot study. Oral Oncol. 1999;35(3):314-320.
13. Feig SA. Screening mammography: a successful public health initiative. Rev Panam Salud Publica. 2006;20(2-3):125-133.
14. Heim K, Widschwendter A, Szedenik H, et al. Specic serologic
response to genital human papillomavirus types in patients with vulvar
precancerous and cancerous lesions. Am J Obstet Gynecol. 2005;192(4):
1073-1083.
15. Quinley KE, Gormley RH, Ratcliffe SJ, et al. Use of mobile telemedicine for cervical cancer screening. J Telemed Telecare. 2011;17(4):203-209.
16. Bhaskaranarayana A, Satyamurthy LS, Remilla ML. Indian Space
Research Organization and telemedicine in India. Telemed J E Health.
2009;15(6):586-591.
17. Klein D, Davis P, Hickey L. Videoconferences for rural physicians
continuing health education. J Telemed Telecare. 2005;11(suppl 1):97-99.
18. Pak HS, Brown-Connolly NE, Bloch C, et al. Global forum on telemedicine: connecting the world through partnerships. Telemed J E Health.
2008;14(4):389-395.
19. Richter GM, Williams SL, Starren J, Flynn JT, Chiang MF. Telemedicine for retinopathy of prematurity diagnosis: evaluation and challenges. Surv Ophthalmol. 2009;54(6):671-685.
20. Hailey D, Roine R, Ohinmaa A. Systematic review of evidence for the
benets of telemedicine. J Telemed Telecare. 2002;8(suppl 1):1-30.
21. Mishra SK, Kapoor L, Singh IP. Telemedicine in India: current scenario and the future. Telemed J E Health. 2009;15(6):568-575.
22. Sankaranarayanan R, Ramadas K, Thara S, et al. Long term effect of
visual screening on oral cancer incidence and mortality in a randomized
trial in Kerala, India. Oral Oncol. 2013;49(4):314-321.
23. Baeten J, Suresh A, Johnson A, et al. Molecular imaging of oral
premalignant and malignant lesions using uorescently labeled lectins.
Transl Oncol. 2014;7(2):213-220.
24. Menon U, Belue R, Wahab S, et al. A randomized trial comparing the
effect of two phone-based interventions on colorectal cancer screening
adherence. Ann Behav Med. 2011;42(3):294-303.
25. Holmstrup P, Vedtofte P, Reibel J, Stoltze K. Long-term treatment
outcome of oral premalignant lesions. Oral Oncol. 2006;42(5):461-474.
26. Sperandio M, Brown AL, Lock C, et al. Predictive value of dysplasia
grading and DNA ploidy in malignant transformation of oral potentially
malignant disorders. Cancer Prev Res (Phila). 2013;6(8):822-831.
27. Warnakulasuriya S, Kovacevic T, Madden P, et al. Factors predicting
malignant transformation in oral potentially malignant disorders among
patients accrued over a 10-year period in South East England. J Oral Pathol
Med. 2011;40(9):677-683.

JADA

( )

- -

http://jada.ada.org

FLA 5.2.0 DTD ! ADAJ173_proof ! 4 July 2015 ! 4:40 pm ! ce

2015 9