The trend of oral cancer screening in Bangladesh, perceptions, and risk factors: a model comparison approach

Irin Yasmin; Md. Shahariar Rokon; Elnaz Illias; Md. Nafiul Alam Khan; Md. Mahbub Alam; Mohammad Delwer Hossain Hawlader; Rajib Kumar Malakar; Faroque Md. Mohsin

doi:10.12688/f1000research.138278.1

Home Browse The trend of oral cancer screening in Bangladesh, perceptions, and...

ALL Metrics

-

Views

-

Downloads

Get PDF

Get XML

Export

▬

✚

Research Article

The trend of oral cancer screening in Bangladesh, perceptions, and risk factors: a model comparison approach

[version 1; peer review: 2 approved with reservations]

Irin Yasmin¹, Md. Shahariar Rokon², Elnaz Illias¹, [...] Md. Nafiul Alam Khan², Md. Mahbub Alam³, Mohammad Delwer Hossain Hawlader¹, Rajib Kumar Malakar⁴, Faroque Md. Mohsin ¹

Irin Yasmin¹, Md. Shahariar Rokon², [...] Elnaz Illias¹, Md. Nafiul Alam Khan², Md. Mahbub Alam³, Mohammad Delwer Hossain Hawlader¹, Rajib Kumar Malakar⁴, Faroque Md. Mohsin ¹

PUBLISHED 25 Jul 2023

Author details Author details

¹ Department of Public Health, North South University, Dhaka, Dhaka Division, Bangladesh
² Department of Statistics, Jahangirnagar University, Savar, Bangladesh
³ Department of Public Health, University of South Asia, Dhaka, Bangladesh
⁴ Bangabandhu Sheikh Muji Medical University, Dhaka, Bangladesh

Irin Yasmin
Roles: Conceptualization, Data Curation, Formal Analysis, Investigation, Methodology, Resources, Writing – Original Draft Preparation, Writing – Review & Editing

Md. Shahariar Rokon
Roles: Conceptualization, Data Curation, Formal Analysis, Investigation, Methodology, Resources, Software, Visualization, Writing – Original Draft Preparation

Elnaz Illias
Roles: Conceptualization, Data Curation, Resources, Validation, Visualization, Writing – Original Draft Preparation, Writing – Review & Editing

Md. Nafiul Alam Khan
Roles: Data Curation, Formal Analysis, Investigation, Methodology, Project Administration, Resources, Software

Md. Mahbub Alam
Roles: Investigation, Project Administration, Resources, Software, Validation, Visualization, Writing – Original Draft Preparation

Mohammad Delwer Hossain Hawlader
Roles: Investigation, Project Administration, Supervision

Rajib Kumar Malakar
Roles: Project Administration, Resources, Software, Validation, Writing – Review & Editing

Faroque Md. Mohsin
Roles: Conceptualization, Methodology, Project Administration, Supervision, Writing – Original Draft Preparation, Writing – Review & Editing

OPEN PEER REVIEW

REVIEWER STATUS

Abstract

Background: Oral cancer (OC) is a leading cause of death among cancer patients and a major public health concern in Bangladesh. Timely OC screening can help in the early detection of OC and could reduce the mortality rate in Bangladesh. We aimed to assess the tend of OC screening, perception and risk factors among dental patients.
Methods: This was a facility-based cross-sectional study among 423 participants with dental problems using a convenience sampling technique. The collected data were analyzed by IBM SPSS 25 and various statistics, such as the chi-squared test, Poisson regression, logistic regression, etc., were used to attain the objectives.
Results: Only 2.4% of the study participants underwent OC screening before the study period. The use of smokeless tobacco and smoking were significantly associated with the tend of OC screening. According to the Poisson model, previous smokeless tobacco users were 5 times (odds ratio (OR): 5.403, P<0.05) more likely than non-users to go for OC screening, which was 8 times (OR: 7.729, P=0.033) for the binary logistic regression. Having a family history of OC increased the odds of receiving an OC screening by nine times (OR: 9.544, P<0.002) compared to individuals without a family history of OC; it was 16 times (OR: 16.438, P=0.001) for the binary logistic regression.
Conclusions: This study shows former smokeless tobacco users and past smokers are more likely to uptake OC screening than non-users. In addition, patients having a family history of OC increased the odds of receiving OC screening than subjects without a family history of OC. The lack of OC screening in Bangladesh highlights the need to implement a national-level screening program to help with the early diagnosis of OC and its prevention.

Keywords

Oral cancer screening, Risk factors, Tobacco, Model comparison approach, Bangladesh

Corresponding author: Faroque Md. Mohsin

Competing interests: No competing interests were disclosed.

Grant information: The author(s) declared that no grants were involved in supporting this work.

Copyright: © 2023 Yasmin I et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

How to cite: Yasmin I, Rokon MS, Illias E et al. The trend of oral cancer screening in Bangladesh, perceptions, and risk factors: a model comparison approach [version 1; peer review: 2 approved with reservations]. F1000Research 2023, 12:882 (https://doi.org/10.12688/f1000research.138278.1) First published: 25 Jul 2023, 12:882 (https://doi.org/10.12688/f1000research.138278.1) Latest published: 25 Jul 2023, 12:882 (https://doi.org/10.12688/f1000research.138278.1)

Introduction

Oral cancer (OC), comprising lip, tongue, and mouth cancers, collectively represents one of the most common cancers worldwide.¹ As with other cancers, OC is a leading cause of death and a barrier to increasing life expectancy globally. A global estimate based on Global Statistics 2020 reports that nearly 377,713 new cases of OC were diagnosed in 185 countries, causing about 177,757 deaths and of which Asian countries accounted for almost two-thirds of all the new cases.² Unfortunately, according to mouth and oral cancer statistics (2020), Bangladesh ranks second in terms of incidences of mouth and OC but first in deaths.³

OC is a multifactorial disease associated with several risk factors including smoking,⁴^–⁶ alcohol,¹^,⁴^,⁶ human papillomavirus,¹^,⁶ smokeless tobacco,⁷^,⁸ family history (first degree family members),⁹ and periodontal disease,¹^,¹⁰ while the smokeless tobacco practice and consumption of tobacco and alcohol are reasons that predispose the Asian population to OC.¹¹ In fact, the prevalence of current tobacco use is 34.7% in Bangladesh,¹² which is particularly worrying because tobacco use is one of the leading causes of OC.

A number of studies suggest that a lack of early diagnosis and screening services, financial constraints, social and cultural factors, and lack of access to cancer services contribute to global disparities in OC outcomes.⁶^,¹¹^,¹³^–¹⁵ OC screening consists of a systematic clinical examination of the oral cavity, which includes examining the face, neck, lips, labial mucosa, buccal mucosa, gingiva, and the floor of the mouth, tongue, and palate.⁴ In order to prevent OC, early detection through screening is highly recommended by several studies,¹⁰^,¹³^,¹⁴ despite some controversy regarding the effectiveness of the assessment.⁶

Detecting potential malignant lesions in the oral cavity can be done either by self-examination or clinical examination, making early detection more accessible and accurate. Subsequently, this can significantly reduce OC diagnostic delays quite often.⁶^,¹³^,¹⁶ Dental providers remain alert for signs of potentially malignant lesions or early-stage cancer in patients during routine oral examinations in practice. Cancer screening produces two special benefits which are “down-stage” and reduced mortality or morbidity. OC is particularly known for its premalignant phase, so the screening process is constructed to not only capture OC but also oral potentially malignant disorders, which are responsible for increasing the risk for OC.¹⁷

A great deal of research on oral cancer screening has been conducted in many places like Europe,¹¹ the USA,⁴ Africa,⁶ and India.¹³ At the same time, little work is being undertaken in Bangladesh despite the highest mortality and significant numbers of OC incidence. As a result, a remarkable improvement in screening rates for OC has not been achieved in Bangladesh. To our knowledge, there is a lack of literature that makes use of different statistical models to predict the variables influencing oral cancer screening.

To mitigate this concern, we can find ways to facilitate OC screening uptake in Bangladesh by exploring the status of OC screening and its potential associated variables. As a result, this study compares various regression models to find the odds of OC screening; those models were previously used in different studies for better understanding and outcomes.¹⁸^–²⁰

Methods

Study design and population

This cross-sectional study was carried out among patients at National Healthcare Networks, Adabor Centre and Anan Specialised Dental Clinics, Uttara, Dhaka, Bangladesh. The inclusion criteria of the participants were (a) patient visited the center with dental problems, (b) age group > 18 years old, (c) willingly agreed to participate in the study.

Sample size

In this study, we used convenience sampling method to collect our desired data. We used this method particularly for the lack of sampling frame, to reduce cost and time. By using formula, we obtained the minimum sample size required for the study.

n_{0} = \frac{Z^{2} pq}{d^{2}}

Where is the sample size, z is the standard normal variate, p is the population’s percentage that possesses the attribute, and p + q = 1, and d is the largest error that can be made while predicting the population percentage. Considering the degrees of accuracy, d = 0.05, z = 1.96 and p = 0.34.¹² Thus, we got the required sample size 342. However, for better understanding and analysis, we targeted more samples to be included. Thus, we gathered 423 unique samples.

Data collection procedures

The data collection was done by the research team. All of them are physicians and directly involved in study centers. Upon receiving written consent, data collectors were recorded patients’ sociodemographic data, clinical history and OC screening information upon inclusion criteria. Moreover, the supervisor randomly checked the collected data forms to avoid errors. For data collection, a semi-structured questionnaire was used that was developed based on previous studies.⁶^,¹⁵ The questionnaire had three parts: the first part described the demographic characteristics of the respondents—the second part of the questionnaire, clinical information and healthcare utilisation and the last part-cancer screening data. Finally, we collected 423 data through face-to-face interviews (response rate: 98%) from the study hospitals between February to March 2022.

Independent and dependent variables

The dependent variable of this study was the presence of OC screening. We categorized the dependent variable into two categories. Which was defined as

Y = \{\begin{cases} 0 & Respondent didn’t opt for oral cancer screening \\ 1 & otherwise \end{cases}

The independent variables of the study were gender, age, marital status, education, occupation, living area, number of family members, monthly income, smokeless tobacco use (paansupari/jorda/tamak pata etc.), smoking status, dentist visit, last visit to the lentist, checkup routine, cost management for treatment, knowledge of OC, medium of getting information about OC, If the respondent had no idea about oral cancer (explain for the lack of knowledge), If OC screening test is free of cost, would the respondent examine or not, family history of OC. We used shortened names of the independent variables for analysis purposes and software restrictions (Table 1).

Table 1. Univariate analysis of the variables.

Variables	Categories	Code	Frequency	Percentages
Response variable
Oral cancer screening (Y)	No	0	410	96.90
Oral cancer screening (Y)	Yes	1	10	2.40
Explanatory variables
Gender	Male	0	203	48.00
Gender	Female	1	220	52.00
Mstatus	Unmarried	0	68	16.10
Mstatus	Married	1	355	83.90
Education	No schooling	0	38	9.00
	Primary	1	36	8.50
	Secondary	2	56	13.20
	Higher secondary	3	88	20.80
	University & more	4	205	48.50
Occupation	Unemployed	0	44	10.40
	Employed	1	151	35.70
	Business	2	51	12.10
	Others	3	177	41.80
Living area	Rural	0	92	21.70
Living area	Urban	1	331	78.30
Tobacco	Never	0	240	56.70
	Current	1	111	26.20
	Past	2	72	17.00
Smoking	Never smoker	0	259	61.20
	Current smoker	1	100	23.60
	Past smoker	2	64	15.10
Visit dentist	Never	0	53	12.50
	< 6 months ago	1	56	13.20
	6-12 months	2	87	20.60
	12 months ago	3	227	53.70
Reason	Regular check up	0	52	12.30
	Occasional	1	59	13.90
	Only when having trouble with teeth	2	312	73.80
Tcost	Personal	0	191	45.20
	Family members	1	165	39.00
	Other sources	2	67	15.80
if yes (media)	Media (TV, Radio, Newspaper)	0	202	47.80
	Internet (Facebook, YouTube, Website)	1	88	20.80
	Family members/relatives/friends	2	64	15.10
	Direct contact with a patient	3	25	5.90
	Others	4	16	3.80
EverOc	No	0	42	9.90
EverOc	Yes	1	381	90.10
If no	I had not dental disease	0	186	44.00
	It’s costly	1	08	1.90
	Not available the facility	2	33	7.80
	Not interested	3	155	36.60
	Others	4	34	8.00
Free test	No	0	128	30.30
Free test	Yes	1	293	69.30
FHO	No	0	407	96.20
FHO	Yes	1	15	3.50
Age	11-20	0	09	2.10
	21-30	1	80	18.90
	31-40	2	109	25.80
	41-50	3	81	19.10
	51-60	4	68	16.10
	61-70	5	64	15.10
	71-80	6	07	1.70
	81-90	7	05	1.20
Monthly income	10-10000	0	289	68.30
	10001-200000	1	108	25.50
	200001-400000	2	19	4.50
	>400000	3	07	1.70
Family members	1-4	0	219	51.80
	5-8	1	192	45.40
	>8	2	12	2.80

Data analysis

All statistical analysis was performed using R (R Core Team, 2020), RStudio (Rstudio Team, 2020), version 2021.9.1.372 and Statistical Packages for Social Science (SPSS) version 26. For background study or univariate analysis, we used frequency distribution. To measure the association between the response and the explanatory variable, we used the chi-square test.²¹ The expression is defined as¹⁰

χ^{2} = \sum_{j = 1}^{n} \frac{{(A_{j} - B_{j})}^{2}}{B_{j}} \dots \dots \dots \dots \dots \dots \dots \dots \dots

Where, A_j observed and B_j expected cell frequencies. Furthermore, the test statistic follows chi-square distribution with (m-1) (n-1) degrees of freedom. Where m is the number of the categories of the covariate and n is the number of responses.

In order to find out the magnitude of the relationship between response and explanatory variable, we conducted three different regression models, namely Poisson regression, Binary logistic regression, Poisson regression with robust variance.

The binary logistic model is generally used for classification problems while utilizing maximum likelihood estimate.²¹ Let a_i(i = 1, 2, …, n) be our outcome variable with categories 1, …, j, …, c and b_i = (b_i1, b_i2, …, b_in)’ as the column vector with k covariates. The expression is defined as²²

log [\frac{\nabla}{1 - \nabla}] = a_{j} + β^{'} b_{i}; for j = 1, 2, \dots, c - 1 \dots \dots \dots \dots \dots .

Where $\nabla$ represents probability of the event happening and $1 - \nabla$ represents probability of it not happening.

Poisson regression²⁰ is frequently used in epidemiology for longitudinal studies in which the outcome variable is the number of episodes of a disease that occur over time. The model formulation is²³

log \frac{n}{t} = β 0 + β 1 X 1 + \dots + βkXk \dots \dots \dots \dots \dots

Where n is the number of counts for given individual over time t. X_i is the model covariates and β_i are the model parameters, also known as log relative risk. Previous studies showed significant improvement over general model using robust variance estimate and we used Poisson regression with robust variance.¹⁸^,¹⁹ Finally, we compared these three models to using standard error and confidence interval of the estimates.

Ethical considerations

The written informed consent was obtained from each participant and the hospital authorities. Participants were assured that their personal information would remain confidential and be used only for academic purposes. In addition, participants were informed that they could withdraw at any time without negative consequences. The study protocol was approved by the Institutional Review Board of North South University in Bangladesh (Ref-2022/OR-NSU/IRB/1004).

Results

For background study, we did frequency analysis to compare their frequencies and percentages, presented in Table 1. This table demonstrated that out of all respondents, 96.9% didn’t opt for OC screening, and only 2.4% had gone through the screening. Among our patients, 52% were females, and 48% were males. 83% of our patients were married, and we had only 16.1% of unmarried patients. Our responders were generally more educated (48.5%) than less educated (38%). Only 12.1% of the patients were self-employed, whereas 35.7% (151) were employed. Only 21.7% of respondents resided in rural areas, while 78.3% lived in cities. Only 17% of the respondents had ever used smokeless tobacco, while 23.6% were current users. More than half of the patients (53.7%) had a dental appointment the previous year compared to only 13.2% who had one within the last six months. Merely 12.3% of respondents reported seeing the dentist on a regular basis; the majority of respondents 73.8% only did so when they experienced dental problems. Most of the participants (96.2%) had no family history of oral cancer, and 68.3% of the patients were from lower-income groups. With a monthly income of more than BDT 400000, only 1.7% (7 of our respondents) were in the higher income group.

Initially we performed chi-square test to determine whether our variables are associated with the dependent variable or not. In Table 2 we showed the results of the chi-square association test.

Table 2. Chi-square test.

Variable	Categories	Oral cancer screening		Chi-square value	p-value
Gender	0	195	5	0.023	0.879
Gender	1	215	5	0.023	0.879
Mstatus	0	67	0	1.944	0.163
Mstatus	1	343	10	1.944	0.163
Education	0	37	1	2.842	0.585
	1	35	1
	2	53	2
	3	88	0
	4	197	6
Occupation	0	42	1	0.958	0.812
	1	147	4
	2	48	2
	3	173	3
Living area	0	89	2	0.017	0.897
Living area	1	321	8	0.017	0.897
Tobacco	0	237	2	9.146	0.010***
	1	107	3
	2	66	5
Smoking	0	255	4	5.255	0.072*
	1	97	2
	2	58	4
Visit dentist	0	52	1	3.367	0.338
	1	54	2
	2	82	4
	3	222	3
Reason	0	48	4	7.991	0.018**
	1	58	0
	2	304	6
Tcost	0	185	3	8.913	0.012**
	1	163	2
	2	62	5
if yes (media)	0	196	3	5.052	0.282
	1	86	2
	2	60	4
	3	24	1
	4	16	0
EverOc	0	41	1	0.000	1.00
EverOc	1	369	9	0.000	1.00
If no	0	182	4	4.614	0.329
	1	08	0
	2	33	0
	3	153	0
	4	33	1
Free test	0	125	2	0.290	0.590
Free test	1	284	7	0.290	0.590
FHO	0	398	6	39.367	0.000***
FHO	1	11	4	39.367	0.000***
Age	0	09	0	8.400	0.299
	1	79	0
	2	104	5
	3	81	0
	4	66	2
	5	59	3
	6	07	0
	7	05	0
Monthly income	0	281	5	2.155	0.541
	1	104	4
	2	18	1
	3	07	0
Family member	0	212	5	0.344	0.842
	1	186	5
	2	12	0

We were able to identify five significant variables that were related to our response variable using the chi-square test. We only selected these five variables to study further.

To determine the optimal result in this research, we applied three alternative regression models. In the first step, we used Poisson regression to find out the estimates as shown in the Table 3.

Table 3. Poisson regression.

Variables	Categories	Estimate	Exp(β)	Std. Error	Pr(>\|z\|)
(Intercept)		-4.4379	0.012	1.0518	2.45e-05***
Tobacco	Never(ref)	-	1	-	-
	Current	0.7874	2.198	1.0311	0.44395
	Past	1.6869	5.403	0.8957	0.05967*
Smoking	Never(ref)	-	1	-	-
	Current	-0.3865	0.679	1.0872	0.72224
	Past	0.8511	2.342	0.8128	0.29506
Reason	Regular Checkup (ref)		1
	Occasional check up	-16.9951	0.000	1883.2278	0.99280
	Only when having trouble with teeth	-1.2070	0.299	0.7789	0.12125
Tcost	Personal income (ref)	-	1	-	-
	Family member	0.2415	1.273	1.0430	0.81687
	Other sources	1.0827	2.953	0.8085	0.18051
FHO	No (ref)	-	1	-	-
FHO	Yes	2.2559	9.544	0.7358	0.00217***

We discovered three statistically significant results in the Poisson regression model. Our model demonstrated that a person will be 98% (Odd Ratio (OR): 0.012, P < 0.000) less likely to opt for OC screening if all other variables remain constant. We also found that tobacco use significantly increased the odds of OC screening. According to the Poisson model, previous tobacco users were 5 (OR: 5.403, P < 0.05) times more likely than non-user to go for OC screening. Finally, having a family history of OC (FHO) increased the odds of OC screening by 9 (OR: 9.544, P < 0.002) times compared to people without a family history of OC.

We employed binary logistic regression to analyze the results in order to conduct further research. As indicated in Table 4 four statistically significant variables were discovered in our research based on the binary logistic regression. If all other variables remain constant, a person had almost no odds (OR: 0.009, P < 0.000) of OC screening. Additionally, binary logistic regression revealed that compared to non-users, previous tobacco use raised the odds of OC screening by a factor of up to 8 (OR: 7.729, P = 0.033). Also, family history of OC raised the odds of OC screening by 16 times (OR: 16.438, P = 0.001) than no family history of OC.

Table 4. Binary logistic regression.

Variables	Categories	Estimate	Exp(β)	Std. Error	Pr(>\|z\|)
(Intercept)		-4.6761	0.009	1.1538	5.06e-05***
Tobacco	Never(ref)	-	1	-	-
	Current	0.7581	2.134	1.0711	0.47906
	Past	2.0449	7.729	0.9633	0.03376**
Smoking	Never(ref)	-	1	-	-
	Current	0.1468	1.158	1.1979	0.90247
	Past	0.9688	2.635	0.8849	0.27356
Reason	Regular checkup (ref)	-	1	-	-
	Occasional check up	-17.6160	0.000	2161.1112	0.99350
	Only when having trouble with teeth	-1.5513	0.212	0.8598	0.07117*
Tcost	Personal income (ref)	-	1	-	-
	Family member	0.5435	1.722	1.1574	0.63866
	Other sources	1.5331	4.632	0.9354	0.10122
FHO	No (ref)	-	1	-	-
FHO	Yes	2.7996	16.438	0.8541	0.00105***

The findings were analyzed using Poisson regression with robust variance, as shown in Table 5. According to our model, the probabilities of going for OC screening were essentially zero if all other factors remained the same (OR: 0.012, P < 0.000). Four additional significant variables were also obtained by the Poisson model with robust variance. When compared to people who had never used tobacco, individuals who had a history of tobacco use had 5 (OR: 5.403, P = 0.05) times higher odds of OC screening. Additionally, former smokers had 2-fold higher likelihood of receiving an OC screening compared to non-users (OR: 2.342; P = 0.08 at the 10% significant level). Furthermore, our data showed that there were no odds of OC screening among those who occasionally visited the dentist for a checkup (Reason 2) (OR: 0.000, P = 0.000). Finally, having a family history of OC increased the odds of OC screening by 9 (OR: 9.544, P < 0.001) times that of those who had no family history of OC. It is evident that the findings of the Poisson regression and the Poisson regression with robust variance were the same, but they differed in the variance, standard error, and confidence interval of the estimates.

Table 5. Robust Poisson regression.

Variables	Categories	Estimate	Exp(β)	Std. Error	Pr(>\|z\|)
(Intercept)		-4.437922	0.012	0.84654	1.585e-07***
Tobacco	Never(ref)	-	1	-	-
	Current	0.78736	2.198	1.10141	0.474692
	Past	1.68691	5.403	0.86698	0.051688*
Smoking	Never(ref)	-	1	-	-
	Current	-0.38645	0.679	1.01648	0.703808
	Past	0.85110	2.342	0.49502	0.085555*
Reason	Regular checkup (ref)	-	1	-	-
	Occasional check up	-16.99510	0.000	0.62012	<2.2e-16***
	Only when having trouble with teeth	-1.20700	0.299	0.81469	0.138461
Tcost	Personal income (ref)	-	1	-	-
	Family member	0.24152	1.273	0.77363	0.754896
	Other sources	1.08273	2.953	0.73953	0.143173
FHO	No (ref)	-	1	-	-
FHO	Yes	2.25590	9.544	0.70389	0.001351***

For better understanding and to find out the optimal model, we compared the odds and confidence interval of the odds ratio as represented in Table 6. The Poisson model produced significantly less standard error and provided narrower confidence intervals than the binary logistic regression, making it clear that it outperformed the latter by a wide margin. However, it becomes clear that the robust Poisson model performed significantly better than both the Poisson and the logistic regression models when compared to the Poisson model with robust variance.

Table 6. Comparison among Poisson, Logistic and Robust Poisson.

Variables	Categories	Poisson		Logistic		Robust Poisson
		odds	CI	odds	CI	odds	CI
(Intercept)	-	0.012	0.002-0.093	0.009	0.001-0.089	0.012	0.002-0.062
Tobacco	Never (ref)	1	-	1	-	1	-
	Current	2.198	0.291-16.583	2.134	0.262-17.416	2.198	0.254-19.031
	Past	5.403	0.934-31.266	7.729	1.170-51.057	5.403	0.988-29.552
Smoking	Never (ref)	1	-	1	-	1	-
	Current	0.679	0.081-5.722	1.158	0.111-12.117	0.679	0.093-4.982
	Past	2.342	0.476-11.521	2.635	0.465-14.927	2.342	0.888-6.18
Reason	Regular Checkup (ref)	1	-	1	-	1	-
	Occasional check up	0.000	-	0.000	-	0.000	-
	Only when having trouble with teeth	0.299	0.065-1.377	0.212	0.039-1.143	0.299	0.061-1.477
Tcost	Personal income (ref)	1	-	1	-	1	-
	Family member	1.273	0.165-9.833	1.722	0.178-16.640	1.273	0.279-5.80
	Other sources	2.953	0.605-14.402	4.632	0.741-28.975	2.953	0.693-12.581
FHO	No (ref)	1	-	1	-	1	-
FHO	Yes	9.544	2.256-40.367	16.438	3.082-87.671	9.544	2.402-37.921

Table 6 shows that the confidence interval for the odds of past tobacco user is 0.934-31.266 for Poisson regression and 1.170-51.057 for logistic regression, but it becomes much narrower for robust Poisson regression at 0.988-29.552, indicating a better estimate interval. Additionally, the Poisson regression’s confidence interval for past “smoking” was 0.476-11.521, whilst the logistic regression’s range was 0.465-14.927. However, robust Poisson regression yielded a significantly narrower interval than those two, at 0.888-6.18. Likewise, the range of the logistic regression was 3.082-87.671, while the range of the Poisson regression’s confidence interval for FHO was 2.256-40.367. However, robust Poisson regression produced an interval that was 2.402-37.921, which was noticeably narrower than those two.

Robust Poisson regression yielded a substantially lower standard error and a much smaller confidence interval for the estimate for each of the relevant examples.

Discussion

Finding from this survey aimed at reporting the tend of OC screening and assessing risk factors associated with ever OC examination. Our findings indicate a lower rate of OC screening, with only 2.4% of respondents indicating that they had ever received OC screening. This finding is lower than the estimated 6.8% in Sudan,⁶ 28% in India,²⁴ and 30% in the USA.²⁵ We found more than half of the patients visited the dentist twelve months ago, and seventy-three percent of patients reported that they visit the dentist when they have trouble with their teeth. These findings give a possible cause for concern as low levels of awareness of oral health may affect the chances of early presentation of OC. A greater emphasis should be placed on patient education regarding OC risk and detection, as there is no population screening program for this condition. A surprising result of this study was that 90% of patients had heard of OC screening. Even though this report indicates that participants are aware of screening for OC, it also exhibits a negative attitude or ignorance of its existence because only 2.4% of participants underwent OC screenings. This is the same situation that has been reported that only a few participants had ever received OC screenings despite hearing about OC screening before in different places.⁶^,¹⁵ This gives cause for concern as low levels of OC screening will affect the chances of early detection of OC. It also highlights that extensive population-based surveys, awareness programs and establishing health literacy may need to be targeted more closely to encourage people to OC screening gradually.

There is a substantial body of evidence linking tobacco use as one of the leading causes of cancer and its death⁵; in that same way, using smokeless tobacco products, such as dipping and chewing tobacco, also leads to cancer mouth, esophagus, and throat cancers.⁸ According to data, compared to people who had never used smokeless tobacco, individuals with a history of smokeless tobacco use had five times higher odds of OC screening. The results of this study support previous studies suggesting that tobacco users had a significantly more positive attitude toward screening than non-tobacco users.¹⁴^,¹⁵ The reason might be that continued anti-tobacco focus strategies such as advertising bans, higher taxes and prices, and restrictions contribute to behavioral changes in Bangladesh, although the country’s implementing smokeless tobacco control policies is insufficient.⁷ This study also observes that similar to the former smokeless tobacco users, former smokers had a 2-fold higher likelihood of receiving an OC screening than non-users which is in stark agreement with a prior study that reported that smoking was a significant determinant of receiving an OC screening.⁴^,¹⁵ On the contrary, previous studies reported that former smokers were not a significant determinant of having a positive attitude or undergoing OC screening.⁴^,¹⁴ A probable explanation for our finding might be that former smokers have a positive attitude toward OC screenings, and therefore, it is likely that opportunistic OC screening during dental checkups will yield positive results.

A previous study indicated that regular visits to dentists or check-ups do not affect the uptake of OC screening.¹⁵ This result in line with our findings that there were no odds of OC screening among those who occasionally visited the dentist for a checkup. This result suggests that using the opportunity a dental appointment may provide to raise awareness may be increasingly vital to encouraging OC screening, and a previous study emphasised it.⁶ However, our study also suggests promoting public awareness concerning OC as a risk factor. The risk of many cancers is higher in subjects with a family history of cancer at a concordant site. It was also evident that OC had increased risks of family history at discordant sites.⁹ This present study demonstrated that having a family history of OC increased the odds of OC screening by nine times that of those with no family history of OC. This result is in line with a previous study conducted in India among women.¹³ The possible reason might be people with a family history of cancer/OC are better aware of cancer as a disease and hence participate more in OC screening.

Of note, we observed a strong influence of the use of smokeless tobacco and smoking in the uptake of OC screening services. As shown in Table 5, the odds of having OC screening were higher among former smokeless tobacco users and former smokers. The findings in the present study concerning current smokeless tobacco users and smokers’ attitudes related to OC suggest strongly that awareness programs for OC screening are indispensable for them. Raising oral cancer awareness during dental appointments may be increasingly crucial to encouraging early detection. While anti-tobacco strategies such as advertising bans, higher taxes and prices, and restrictions continue to be implemented in Bangladesh, smokeless tobacco control policies remain inadequate.⁷ Therefore, policymakers must evolve and introduce new effective smokeless tobacco control policies. Additionally, this study revealed that visits to dentists or check-ups do not affect the uptake of OC screening. It concerns that low awareness among people and dentists is likely to hinder early presentation. To address this barrier, dental professionals and patients may need educational initiatives like training programs to improve health literacy surrounding OC risk factors. As well, general dentists’ confidence, expertise, and knowledge in conducting future OC screenings may need to be improved by implementing theory-based interventions.

Strength and limitations

To our knowledge, this study is the first to assess the tend and associated factors of OC screening among dental patients in Bangladesh. As well this is the first paper to compare OC screening ratios and confidence intervals from different strategies to a suitable reference. In this study, we particularly examined the odds ratio produced by several models. This research is among the first to compare different regression models while comparing their findings about OC screening in Bangladesh. Although qualitative studies are prevalent in OC screening, model comparison and model selection are less focused. So, we presented a different point of view when discussing OC screening. In order to initiate early treatment and avoid OC in its later stages, we used a variety of parameters in this study to determine the likelihood of OC screening. This research also emphasises already established factors (i.e., smoking and using tobacco) responsible for OC and further interpret the findings for the OC screening scenario in Bangladesh. Due to the nature of the study, we employed non-probability sampling for procuring and extracting the findings.

While this research makes several contributions to understanding OC screening, there are some limitations. In this study, we used convenience sampling, which is a non-probability sampling method. So, the samples are not chosen by random selection. We collected samples that were readily available to us. So, it is only possible to generalise for some of the population. Due to the nature of the sampling method, the research may be subject to under-coverage bias and observer bias.

Conclusion

This study disclosed the tend and associated risk factors of OC screening among Bangladeshi dental patients using a model comparison approach. This study has found that only 2.4% of dental patients underwent OC screening before the study period. We also have found that former smokeless tobacco users and past smokers are more likely to uptake OC screening than non-users. In addition, patients having a family history of OC increased the odds of receiving OC screening than subjects without a family history of OC. The lack of OC screening in Bangladesh highlights the need to implement a national-level screening program to help with the early diagnosis of OC and its prevention.

Data availability

Underlying data

Open Science Framework. Association between Mediterranean diet adherence and dyslipidemia among diabetes mellitus patients. DOI: https://osf.io/2zh6y/.

This project contains the following underlying data:

- data.xlsx (Complete survey responses)

References

1. Oral health: World Health Organization (WHO). Accessed 17 Dec 2022. Reference Source
2. Sung H, Ferlay J, Siegel RL, et al.: Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. 2021; 71: 209–249. PubMed Abstract | Publisher Full Text
3. Mouth and oral cancer statistics|WCRF International: Accessed 17 Dec 2022. Reference Source
4. Akinkugbe AA, Garcia DT, Brickhouse TH, et al.: Lifestyle risk factor related disparities in oral cancer examination in the U.S: A population-based cross-sectional study. BMC Public Health. 2020; 20: 1–11. Publisher Full Text
5. Islami F, Ann GS, Miller KD, et al.: Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. CA Cancer J. Clin. 2018; 68: 31–54. PubMed Abstract | Publisher Full Text
6. Babiker TM, Osman KAA, Mohamed SA, et al.: Oral Cancer Awareness Among Dental Patients in Omdurman, Sudan: a cross-sectional Study. BMC Oral Health. 2017; 17: 69. PubMed Abstract | Publisher Full Text | Free Full Text
7. Naznin E, Wynne O, George J, et al.: Smokeless tobacco policy in Bangladesh: A stakeholder study of compatibility with the World Health Organization’s Framework Convention on Tobacco Control. Drug Alcohol Rev. 2021; 40: 856–863. PubMed Abstract | Publisher Full Text
8. Mu G, Wang J, Liu Z, et al.: Association between smokeless tobacco use and oral cavity cancer risk in women compared with men: a systematic review and meta-analysis. BMC Cancer. 2021; 21: 960. PubMed Abstract | Publisher Full Text | Free Full Text
9. Turati F, Edefonti V, Bosetti C, et al.: Family history of cancer and the risk of cancer: A network of case-control studies. Ann. Oncol. 2013; 24: 2651–2656. PubMed Abstract | Publisher Full Text
10. Lingen MW, Kalmar JR, Karrison T, et al.: Critical evaluation of diagnostic aids for the detection of oral cancer. Oral Oncol. 2008; 44: 10–22. PubMed Abstract | Publisher Full Text | Free Full Text
11. Saraswat N, Pillay R, Everett B, et al.: Knowledge, attitudes and practices of South Asian immigrants in developed countries regarding oral cancer: An integrative review. BMC Cancer. 2020; 20: 1–16. Publisher Full Text
12. Prevalence of current tobacco use (% of adults) - Bangladesh|Data: Accessed 16 Dec 2022. Reference Source
13. Mishra GA, Shaikh HA, Pimple SA, et al.: Determinants of compliance to population-based oral cancer screening program among low socioeconomic women in Mumbai, India. Indian J. Community Med. 2021; 46: 210–215. PubMed Abstract | Publisher Full Text
14. Netuveli G, Sheiham A, Watt RG: Does the “inverse screening law” apply to oral cancer screening and regular dental check-ups? J. Med. Screen. 2016; 13: 47–50. PubMed Abstract | Publisher Full Text
15. Awojobi O, Scott SE, Newton T: Patients’ perceptions of oral cancer screening in dental practice: A cross-sectional study. BMC Oral Health. 2012; 12: 1–9. Publisher Full Text
16. Bouvard V, Nethan ST, Singh D, et al.: IARC Perspective on Oral Cancer Prevention. N. Engl. J. Med. 2022 Nov 24; 387(21): 1999–2005. PubMed Abstract | Publisher Full Text
17. Warnakulasuriya S, Kujan O, Aguirre-Urizar JM, et al.: Oral potentially malignant disorders: A consensus report from an international seminar on nomenclature and classification, convened by the WHO Collaborating Centre for Oral Cancer. Oral Dis. 2021; 27: 1862–1880. PubMed Abstract | Publisher Full Text
18. Lin DY, Wei LJ: The robust inference for the cox proportional hazards model. J. Am. Stat. Assoc. 1989; 84: 1074–1078. Publisher Full Text
19. Barros AJD, Hirakata VN: Alternatives for logistic regression in cross-sectional studies: An empirical comparison of models that directly estimate the prevalence ratio. BMC Med. Res. Methodol. 2003; 3: 1–13.
20. Régression logistique vs autres modèles linéaires généralisés pour l’estimation de rapports de prévalences. Revue d’Epidémiologie et de Santé Publique. 1999; 47: 593–604. PubMed Abstract
21. Pearson K: Contributions to the mathematical theory of evolution. J. R. Stat. Soc. 1893 Dec 1; 56(4): 675–679.
22. WHO: Pathology and Genetics of Head and Neck Tumours. IARC Press; 2005. Accessed 16 Dec 2022. Reference Source
23. Ghantous Y, Abu EI: GLOBAL INCIDENCE AND RISK FACTORS OF ORAL CANCER. Harefuah. 2017 Oct; 156(10): 645–649. PubMed Abstract
24. Matos S, Boakye EA, Crosby D, et al.: Prevalence and Factors Associated with Oral Cavity and Pharyngeal Cancer Screening in a Rural Population. OTO Open. 2021; 5: 1–7. Publisher Full Text
25. Kravietz A, Angara P, Le M, et al.: Disparities in Screening for Head and Neck Cancer: Evidence from the NHANES, 2011-2014. Otolaryngol. Head Neck Surg. 2018; 159: 683–691. Publisher Full Text

Comments on this article Comments (0)

Version 1

VERSION 1 PUBLISHED 25 Jul 2023