Smokeless tobacco use and oral potentially malignant disorders among people living with HIV (PLHIV) in Pune, India: Implications for oral cancer screening in PLHIV

Ivan Marbaniang; Samir Joshi; Shashikala Sangle; Samir Khaire; Rahul Thakur; Amol Chavan; Nikhil Gupte; Vandana Kulkarni; Prasad Deshpande; Smita Nimkar; Vidya Mave

doi:10.1371/journal.pone.0270876

. 2022 Jul 5;17(7):e0270876. doi: 10.1371/journal.pone.0270876

Smokeless tobacco use and oral potentially malignant disorders among people living with HIV (PLHIV) in Pune, India: Implications for oral cancer screening in PLHIV

Ivan Marbaniang ^1,^2,^*, Samir Joshi ³, Shashikala Sangle ⁴, Samir Khaire ⁵, Rahul Thakur ³, Amol Chavan ¹, Nikhil Gupte ^1,⁶, Vandana Kulkarni ¹, Prasad Deshpande ¹, Smita Nimkar ¹, Vidya Mave ^1,⁶

Editor: Jitendra Kumar Meena⁷

PMCID: PMC9255739 PMID: 35788753

Abstract

Introduction

In India, smokeless tobacco (SLT) is a predominant form of tobacco used among people living with HIV (PLHIV). Despite SLT being a risk factor for oral potentially malignant disorders (OPMDs), no prior studies have quantified the association of OPMDs with SLT use among PLHIV. This limits the planning of preventive and control strategies for oral cancer among PLHIV, who are at higher risk for the disease.

Methods

We enrolled 601 PLHIV and 633 HIV-uninfected individuals in an oral cancer screening study at BJ Government Medical College, Pune, India. Oral cavity images were collected using an m-Health application and reviewed by three clinicians. Participants with two clinician positive diagnoses were deemed to have suspected OPMDs. Prevalence ratios (PRs) were used to quantify the association between suspected OPMDs and SLT use among PLHIV. PRs for current SLT users, across HIV status and use duration were also estimated. Corrected PRs were obtained by modifying the maximum likelihood estimation. Models were adjusted for age, smoking, alcohol use and CD4 counts.

Results

Of those enrolled, 61% were men, median age was 36 years (IQR: 28–44), and 33% currently use SLT. Proportion of current SLT users was similar across PLHIV and HIV-uninfected groups but use duration for current SLT use was higher among PLHIV(p<0.05). Among PLHIV, current SLT users had a 5-times (95% CI:3.1–7.0) higher prevalence of suspected OPMDs, compared to non-users. Relative to HIV uninfected individuals with the same SLT use duration, significant associations with suspected OPMDs were seen for PLHIV with<10 use years (PR: 3.5, 95% CI: 1.5–8.1) but not for PLHIV with≥10 use years (PR: 1.3, 95% CI: 0.9–1.8).

Conclusion

PLHIV that are current SLT users are at high risk of OPMDs and potentially oral cancer. The development of strategies for screening, early detection, and management of OPMDs must be considered for this group.

Introduction

In 2020, India contributed to more than a third of the global incident cases and deaths related to oral cancer [1]. People living with HIV (PLHIV) have a higher risk of oral cancer [2, 3], and in India this risk is estimated to be as high as 27 times that of the general population [4]. Despite this higher risk, HIV has not been included as a risk factor for oral cancer in the Indian government’s operational framework for cancer management [5], and oral cancer preventive measures for PLHIV continue to be lacking.

Oral cancers are often preceded by potentially malignant disorders (OPMDs), which include several premalignant conditions with varying potentials for malignant transformation [6]. The ease of access of the oral cavity without privacy requirements, makes OPMDs readily amenable to detection even in resource limited settings [7]. Moreover, as the estimated average time to malignant transformation is 4–10 years, OPMDs present an important phase of opportunity to introduce interventions that disrupt the natural history of oral cancer [8, 9]. Prevalence estimates of OPMDs for Indian HIV uninfected individuals are among the highest in the world [9, 10]. Although an even higher prevalence of OPMDs is implicated for Indian PLHIV by the higher reported risk of oral cancer [4], no prior studies have quantified this supposition or tested for its validity. To inform the development of oral cancer preventive measures for Indian PLHIV, it is crucial to address this research gap.

According to the 2016–2017 Global Adult Tobacco Survey (GATS-2), the prevalence of adult smokeless tobacco (SLT) use in India is 21.4% (29.6% in men and 12.8% in women) [11]. Due to its cultural acceptability across genders and high prevalence of use, SLT is a predominant risk factor for OPMDs in India [6, 9]. For the Indian general population, SLT use is estimated to be associated with 15-times higher odds of OPMDs [6]. For PLHIV, results from a meta-analysis that used data from the Demographic and Health Survey (DHS) which included 18,224 individuals residing in 28 low- and middle-income countries (LMICs), indicate that the prevalence of SLT use could be 1.3 times higher compared to the general population [12]. However, it is currently not known if the magnitude of the association between OPMDs and SLT use in PLHIV is comparable to the one observed for the general population. This makes it challenging to infer if SLT control strategies should be formulated differently for PLHIV, from those for the general population. SLT use is also only assessed as any (lifetime) or no use in many previous studies [6]. However, as the risk of OPMDs reduces considerably after SLT cessation [13], understanding the association between OPMDs and SLT use specifically among current users while accounting for duration of use, could generate data that is better suited for the development of SLT control strategies that are more resource efficient.

In this study, leveraging an m-Health approach, we sought to compare the prevalence of OPMDs among PLHIV to HIV uninfected individuals; quantify the association of OPMDs with SLT use among PLHIV; and estimate the association between OPMDs and SLT use for current users by use duration, comparing PLHIV to HIV uninfected individuals.

Methods

Study setting and design

We used data from an oral cancer screening study conducted at Byramjee Jeejeebhoy Government Medical College and Sassoon General Hospitals (BJGMC-SGH), Pune, India. BJGMC-SGH is a publicly funded tertiary care teaching hospital. It primarily caters to low and lower-middle socioeconomic groups of individuals from the surrounding urban and peri-urban areas of Pune city. Approximately 5000 PLHIV are in active follow-up at the affiliated antiretroviral therapy (ART) center. The annual turnovers of patients at the Otorhinolaryngology and Internal Medicine outpatient clinics are roughly 26,000 and 46,000, respectively.

To calculate the sample sizes for PLHIV and HIV-uninfected participants to be enrolled, we made the following assumptions: 1) The prevalence of OPMDs among HIV-uninfected individuals was assumed to be 10.5%, as reported in a meta-analysis for Asian populations [10]; 2) The prevalence ratio for OPMDs was assumed to be 1.5 among PLHIV compared to HIV-uninfected individuals. Fixing a two-sided confidence interval at 95%, and an enrolment ratio of 1:1 (between HIV-uninfected individuals and PLHIV), we expected to enroll 1320 participants overall (660 PLHIV and 660 HIV-uninfected participants). Under these conditions, we were 80% powered to observe an OPMD prevalence of 15.7% among PLHIV. However, as the study proceeded, due to budgetary constraints, we were unable to meet the originally targeted enrolments. Reducing the power to 77% to detect the same prevalence, the required number of participants was revised to 1226 (613 PLHIV and 613 HIV-uninfected). Power calculations were performed using Epi Info^™ StatCalc (Centers for Disease Control—CDC, USA).

PLHIV attending the ART center and HIV uninfected participants from the outpatient clinics of Otorhinolaryngology and Internal Medicine were recruited by convenience sampling. Eligibility criteria included age≥21 years, absence of oral cancer history, and willingness to provide a written informed consent in Marathi, Hindi (locally spoken languages) or English. HIV uninfected individuals had to additionally consent for a rapid HIV test. All PLHIV were on first or second-line ART regimens as recommended by India’s national HIV control program. Participants were enrolled into the study between June 2017 and June 2019.

The Institutional Review Board of Johns Hopkins University (JHU) and Ethics Committee of BJGMC-JHU Clinical Research Site approved the project.

Study procedures

The study procedures described below were first piloted for 20 participants, the data from whom have not been included in subsequent analyses.

m-Health procedures

An m-health application jointly developed by investigators at JHU and Boston University as a tool for oral cancer screening was used. The application is designed to populate pre-programmed electronic forms and enable the capture of images. It has been extensively tested in different locations in India [14, 15]. Two trained non-medical health care workers entered participants’ information on SLT, smoked tobacco and alcohol use along with sociodemographic, clinical, and sexual history details directly on the application. Eight images of different sites of participants’ oral cavities, where OPMDs are commonly diagnosed, were also obtained by health care workers, using an 8-megapixel phone camera. These included: 1) right gingivobuccal mucosa and retromolar trigone; 2) left gingivobuccal mucosa and retromolar trigone; 3) lower and upper labial mucosa; 4) palate; 5) dorsum of the tongue; 6) floor of mouth and ventral surface of the tongue; 7) right lateral border of the tongue; and 8) left lateral border of the tongue. Data (information and images) were then synced onto a secure cloud-based server.

Oral Human Papilloma Virus (HPV) procedures

Participants also provided oral rinse and gargle samples. Detection of oral HPV was performed by using an in-house real-time HPV PCR (GenePath Diagnostics, Pune), and positive samples were sequenced on Illumina MiSeq. Due to technical issues, 230 samples (19 PLHIV, 211 HIV uninfected) were not processed for oral HPV.

Images review procedures

Images were accessed and reviewed first by two clinicians (a maxillofacial & oral surgeon and an otorhinolaryngologist), each with at least five years of clinical experience in Head and Neck conditions. Clinicians were provided the age, sex, and oral symptoms (negative and positive) including duration of symptoms (if any) but blinded to the HIV status of participants. No SLT, smoked tobacco or alcohol use details were provided. After reviewing each participant’s set of 8 images, depending on whether the clinician considered the images suggestive of having any OPMDs or not, each clinician independently issued a negative or positive ‘presumptive OPMD’ diagnosis. Clinicians’ responses were then reviewed by a study coordinator weekly. If the clinicians disagreed, a senior otorhinolaryngologist with >15 years of clinical experience in Head and Neck oncological surgery served as adjudicator. No prior structured training program to identify suspicious lesions was provided to the any of the clinicians. Participants with two negative presumptive OPMD diagnoses were provided telephonic oral health counselling. Participants with two positive presumptive OPMD diagnoses were asked to come for in-person clinical examinations.

Follow-up procedures

Health care workers made up to 5 attempts to call participants with a) two positive presumptive oral HPV diagnoses b) those with a positive oral HPV result. When successfully contacted, participants were instructed to come for an in-person follow-up and their visits were scheduled. For participants that only required telephonic oral health counselling, up to 2 contact attempts were made. All calls were made within two weeks of images review or oral HPV results. Participants successfully linked to clinicians, had relevant clinical examinations performed and were managed according to BGJMC-SGH guidelines.

A scheme of the study procedures is provided in Fig 1.

Study definitions

Suspected OPMD (outcome variable)

Classified as positive or negative. A participant with two positive presumptive diagnoses was classified as ‘suspected OPMD positive’; a participant with two negative presumptive diagnoses was classified as ‘suspected OPMD negative’.

SLT use (explanatory variable)

Categorized as never, former, and current users, consistent with prior studies [16]. Duration of use for current SLT users was categorized into two groups, <10 and ≥10 years, based on the median use duration among HIV uninfected participants.

Statistical analysis

Medians and proportions were compared using Wilcoxson rank sum and Fisher’s exact tests, respectively. Agreement between clinicians was calculated using the kappa statistic. Modified Poisson regression models with robust estimation of standard errors were fit to estimate prevalence ratios (PRs). First, the PR of suspected OPMDs comparing PLHIV to HIV uninfected individuals was estimated. Second, the analysis was restricted to PLHIV and PRs of suspected OPMDs were estimated across categories of SLT use. Last, PRs of suspected OPMDs were estimated exclusively for current SLT users by HIV status and use duration. Covariates in multivariable models included variables identified a priori to be associated with OPMDs or that were significantly associated (p<0.05) with suspected OPMDs in univariate models. Covariates included age, smoked tobacco use, alcohol use, and time-updated CD4 counts (only for models restricted to PLHIV). To prevent extrapolation by the model, sex was excluded as a covariate as there were ≤5 women that currently smoke or use alcohol.

We performed three sensitivity analyses to support our primary analyses. 1) We found that the kappa statistic between clinicians ranged between 0.16 to 0.33, indicating poor to slight agreement. Given this low inter-clinician agreement in outcome determination, we posited that the misclassification of the outcome variable would be high (i.e., participants who should have been classified as suspected OPMDs were not and vice versa). Also, only 46% of the participants complied with follow-up procedures. Therefore, we were further unable to validate images review diagnoses with in-person clinical findings. We considered the low compliance of participants with follow-up procedures to be of greater concern than the low agreement between clinicians. Put another way, even if the inter-clinician reliability would have been high (i.e., kappa statistic approaching 1), estimates of prevalence and PRs based exclusively on suspected OPMDs would be less relevant clinically, unless these were validated by in-person examinations. Therefore, we viewed our distribution of the outcome variable (suspected OPMDs) obtained under low inter-clinician reliability as arising from a dataset where the outcome is misclassified relative to in-person clinical examinations (i.e., participants that would have been deemed to have OPMD on clinical examination were not and vice versa). To address this outcome misclassification, we carried out the following procedures: a) validity parameters reported in literature comparing remote (images) to in-person diagnosis of oral conditions were identified [16–18]; b) these parameters were then used to generate ‘corrected prevalence’ estimates for OPMDs i.e., prevalence estimates for clinical OPMDs in our population, using the formulae specified by Lash et al. [19]. We excluded validity parameters that generated negative estimates; c) validity parameters thus identified were also used to modify the maximum likelihood estimation to obtain ‘corrected’ odds ratios, adjusted for covariates, as described by Neuhaus, Lyles and Lin, and Lyles et al. [20, 21]. The adjusted odds ratios were then transformed to PRs. Procedures b) and c) were performed assuming non-differential misclassification error of the outcome variable as validity parameters generated external to the study were used, and clinicians made images review diagnoses independent of SLT use and HIV status knowledge. The range of corrected estimates generated through this sensitivity analysis could thus be more clinically meaningful, as they allowed us to infer what the prevalence and PRs could have been under high participant follow-up conditions. Further, conventionally non-differential misclassification of the outcome has mostly been qualitatively described to bias associations towards the null [22]. Estimation of covariate adjusted corrected associations for non-differential misclassification of the outcome variable is also uncommon [21]. Through this sensitivity analysis, we quantitatively estimated corrected adjusted PRs. For a detailed description of the methodology used to obtain corrected prevalence and PRs, please refer to the supplementary files. 2) Since missing data for oral HPV status (a proposed covariate) was high, PRs were compared with multivariable models in which imputed HPV status values estimated using chained equations were included. Results from models excluding oral HPV status as a covariate were treated as primary findings if estimates from the imputed models were comparable. 3) We compared estimates between models in which sex was excluded as a covariate to those in which it was included.

All analyses were performed using Stata 16.1.

Results

Study population

A total of 3112 potential participants were approached (1772 PLHIV, 1340 HIV uninfected individuals). Of these 1234 consented to be enrolled, including 601 PLHIV and 633 HIV uninfected individuals. Males constituted 61% of the study population, and 66% were ≤40 years of age. Thirty-nine percent had ever used SLT, of which 33% were current users. Among current users, the median duration of SLT use was 12 years (IQR: 5–20). The proportions of current alcohol and smoked tobacco users were 26% and 11%, respectively. Oral HPV was detected in 6% (3% oncogenic HPV), and 11% reported ever having performed oral sex. Median time-updated CD4 counts for PLHIV were 508 cells/mm³ (IQR:347–700).

The proportion of current SLT users was similar between PLHIV and HIV uninfected groups (33%). However, SLT use duration among current users was significantly higher among PLHIV than HIV uninfected individuals (PLHIV median SLT use duration:15 years, IQR: 10–20; versus HIV-uninfected SLT median use duration: 10 years, IQR: 4–15, p<0.05) (Table 1).

Table 1. Demographic and clinical characteristics of participants recruited in an mHealth-based oral cancer screening study in Pune, India.

	Overall N (%) / Median (IQR)	PLHIV N (%) / Median (IQR)	HIV uninfected N (%) / Median (IQR)
N (%)	1234	601	633
Suspected OPMDs
No	1048 (85)	484 (81)	564 (89)
Yes	186 (15)	117 (19)	69 (11)
Human Papilloma Virus ^▯
Negative	945 (94)	536 (92)	409 (97)
Positive	59 (6)	46 (8)	13 (3)
Age (years) ^▯
≤ 30	399 (32)	94 (16)	305 (48)
31–40	414 (34)	226 (38)	188 (30)
41–50	312 (25)	219 (36)	93 (15)
> 50	109 (9)	62 (10)	47 (7)
Sex ^▯
Male	753 (61)	320 (53)	433 (68)
Female	481 (39)	281 (47)	200 (32)
Smoked tobacco use ^▯
Never	980 (79)	491 (81)	489 (77)
Former	122 (10)	67 (11)	55 (9)
Current	130 (11)	42 (7)	88 (14)
Number of cigarettes/bidis per day
(current tobacco smoker)	2 (1–3)	2 (1–4)	2 (1–3)
Duration of smoked tobacco use ^▯
(years) (current smoked tobacco users)	5 (2–10)	9 (5–12)	4 (2–8)
Smokeless tobacco use ^▯
Never	751 (61)	353 (59)	398 (63)
Former	70 (6)	47 (8)	23 (4)
Current	413 (33)	201 (33)	212 (33)
Duration of smokeless tobacco use ^▯
(years) (current smokeless tobacco users)	12 (5–20)	15 (10–20)	10 (4–15)
Alcohol use ^▯
Never	766 (62)	390 (65)	376 (59)
Former	144 (12)	88 (15)	56 (9)
Current	324 (26)	123 (20)	201 (32)
Duration of alcohol use ^▯
(years) (current alcohol users)	7 (4–15)	10 (5–15)	5 (3–10)
Multiple sexual partners ^▯
No	957 (78)	445 (74)	512 (81)
Yes	276 (22)	155 (26)	121 (19)
Oral sex
No	1093 (89)	538 (90)	555 (88)
Yes	141 (11)	63 (10)	78 (12)
HIV diagnosis duration (years)
< 5	-	215 (36)	-
5–10		191 (32)
> 10		195 (32)
Median first CD4 counts (cells/mm³⁾	-	228 (120–464)	-
Median time-updated CD4 counts (cells/mm³)	-	508 (347–700)	-

Open in a new tab

Smokeless tobacco use: use of tobacco forms that are not burnt. Locally available forms are khaini (tobacco + slaked lime); gutka (tobacco + areca nut + slaked lime); mishri (roasted powdered tobacco); paan (tobacco + areca nut + slaked lime + condiments, wrapped in a betel leaf); paan masala; snuff.

Human Papilloma Virus types included: HPV 16,18,31,33,35,39,45,51,52,56,58,59.

Multiple sexual partners ≥1 lifetime sexual partner; Oral sex: performed oral sex in their lifetime.

^▯ p-value <0.05 between medians or proportions comparing PLHIV and HIV uninfected.

There are 230 missing values for oral HPV, 211 of which are for the HIV uninfected group, missingness for all other variables is <5%, except first CD4 count (missingness 26.3%).

Prevalence of suspected and corrected OPMDs

Clinicians reviewed a total of 9,872 images, of which 98% were deemed of good quality to make a diagnosis, and the remaining 2% sufficient to reach a diagnosis. The prevalence of suspected OPMDs was 15% (n = 186) for the entire study population, 19% (n = 117) for PLHIV and 11% (n = 69) for the HIV uninfected group. Depending on the extent of suspected OPMDs misclassification, corrected prevalence estimates ranged between 11.6%–37.0% for the entire study population, 16.3%–48.7% for PLHIV and 6.8%–24.5% for the HIV uninfected group (Fig 2; S1 Fig in S1 File).

Associations of suspected and corrected OPMDs with HIV status and SLT use

In the unadjusted model, relative to HIV uninfected individuals, the prevalence of suspected OPMDs among PLHIV was 1.79 (95% CI: 1.36–2.35) times higher. When adjusted for covariates, the association remained statistically significant, albeit reduced in magnitude (PR:1.47, 95% CI: 1.11–1.96) (Table 2, Model 1). Corrected adjusted PRs for the association between HIV positive status and OPMDs ranged between 1.38 to 1.64 (95% CI lower limit range: 1.05–1.14, upper limit range 1.83–2.37) (Fig 3; S2a Fig in S1 File).

Table 2. Prevalence ratios for suspected oral potentially malignant disorders among participants recruited in an mHealth-based oral cancer screening study in Pune, India.

	Suspected OPMDs n (%)	Unadjusted Prevalence Ratio (95% CI)	Adjusted Prevalence Ratio ^ⱡ (95% CI)
Model 1: Total study population (n = 1234)
HIV status
Uninfected	69 (11)	Ref	-
Living with HIV	117 (19)	1.79 (1.36–2.35)	1.47 (1.11–1.96)
Model 2: People living with HIV (n = 601) ^▯
Smokeless tobacco use
Never	30 (8)	Ref	Ref
Former	9 (19)	2.25 (1.14–4.45)	2.40 (0.96–4.34)
Current	78 (39)	4.57 (3.11–6.70)	4.63 (3.06–7.01)
Smoked tobacco use
Never	89 (18)	Ref	Ref
Former	17 (25)	1.40 (0.89–2.20)	0.88 (0.56–1.39)
Current	11 (26)	1.44 (0.84–2.48)	1.08 (0.61–1.91)
Alcohol use
Never	60 (15)	Ref	Ref
Former	22 (25)	1.63 (1.06–2.50)	0.85 (0.54–1.36)
Current	35 (28)	1.85 (1.28–2.66)	0.95 (0.66–1.37)
Age/5 (years)	-	1.16 (1.07–1.26)	1.10 (1.01–1.21)
Sex
Male	72 (22)	Ref
Female	45 (16)	0.72 (0.51–1)	-
Human Papilloma Virus
Negative	106 (20)	Ref	-
Positive	6 (13)	0.66 (0.31–1.42)
Multiple sexual partners
No	79 (18)	Ref	-
Yes	38 (24)	1.38 (0.98–1.94)
Oral sex
No	104 (19)	Ref	-
Yes	13 (21)	1.07 (0.64–1.79)
Time-updated CD4/ 50 (cells/mm ³ )	-	0.99 (0.96–1.03)	1.01 (0.98–1.04)
HIV diagnosis duration (years)	-	0.99 (0.97–1.02)	-

Open in a new tab

^▯ Percentages in parentheses for Model 2 are derived by dividing the frequency observed for OPMD in that category by the total number of PLHIV in those categories as presented in Table 1.

^ⱡ Model 1 was adjusted for age, smoked tobacco use, smokeless tobacco use, and alcohol use. The same variables were adjusted for in Model 2 with CD4 as an additional covariate.

When we restricted the analysis to PLHIV, former (PR: 2.26, 95% CI:1.14–4.45) and current (PR: 4.57, 95% CI 3.11–6.70) SLT users had a higher prevalence of suspected OPMDs compared to never users, in the unadjusted model. When adjusted for covariates, the association for current (PR: 4.63, 95% CI: 3.06–7.01) but not former (PR: 2.04, 95% CI: 0.96–4.34) SLT users remained statistically significant (Table 2, Model 2). Corrected adjusted PRs for current users ranged between 4.82–8.88 (95% CI lower limit range: 3.01–3.99, upper limit range: 7.69–21.42) (Fig 3; S2b Fig in S1 File).

Among current SLT users, compared to HIV uninfected individuals and stratified by use duration, PLHIV with <10 years of use had a higher adjusted prevalence of suspected OPMDs (PR: 3.46, 95% CI: 1.48–8.09). A non-statistically significant association was seen for PLHIV with ≥10 years of use (PR: 1.28, 95% CI: 0.90–1.81) (Fig 4). The range of corrected adjusted PRs for PLHIV<10 years of use was 3.35–12.37 (95% CI lower limit range: 1.46–2.48, upper limit range: 7.72–61.63), and for PLHIV≥10 years of use was 1.27–1.39 (95% CI lower limit range: 0.76–0.91, upper limit range: 1.76–1.95).

Findings from sensitivity analyses indicate that the addition of sex or imputed values of HPV as covariates did not affect our primary findings. However, former SLT use (PR: 2.20, 95% CI: 1.03–4.69) became significantly associated with suspected OPMDs in the model that additionally adjusted for sex. This finding is most likely an extrapolation by the model as discussed earlier (S1 and S2 Tables and S3 Fig in S1 File).

Discussion

In this paper, using data from an m-Health based oral cancer screening study, we estimated the prevalence of OPMDs in Indian PLHIV and compared it with the prevalence in HIV uninfected individuals. Additionally, to our knowledge, we quantified the associations between SLT use and OPMDs in PLHIV for the first time and obtained more granular estimates for the associations between current SLT use and OPMDs by use duration and HIV status. We found that PLHIV were 4%–6% more likely to have OPMDs compared to HIV uninfected individuals, consistent with the known relationship between HIV seropositivity and oral cancer risk [2, 3]. Among PLHIV, relative to non-users, the prevalence of OPMDs was 5–9 times higher among current SLT users. Taken together with the higher prevalence of OPMDs observed among PLHIV, our results indicate that PLHIV that are current SLT users could have the highest risk of OPMDs of any group. Our findings highlight the need to plan oral cancer preventive measures for PLHIV, especially for current SLT users.

Cessation of SLT use among PLHIV that are current users would require behavioral modification and as 49% of SLT users in India express no desire to quit [11], primary preventive measures might be challenging to implement. In this context, screening, early detection, and management of OPMDs could be more practicable as a preventive strategy. Preliminary data from Taiwan suggests that screening for OPMDs can be successfully integrated into primary care [23]. A similar approach culturally adapted for India and integrated into routine HIV care could greatly benefit PLHIV that are current SLT users. Oral cancer screening targeted at high-risk groups has been shown to reduce 15-year oral cancer incidence and mortality by 38% and 81%, respectively [24]. Additionally, findings from a recent study indicate that screening targeted at the highest risk group improves resource efficiency without compromising diagnostic sensitivity [25]. Given India’s low spending on public health and its overburdened health system [26], our results provide preliminary evidence to support the prioritization of PLHIV that are current SLT users, when oral cancer screening programs for PLHIV are formulated to optimize resource allocation.

Among current SLT users with <10 years of use, PLHIV were found to have a significantly higher prevalence of OPMDs compared to HIV uninfected individuals, consistent with our previous findings. However, we did not observe a significant association between OPMDs and HIV seropositivity when SLT use duration was ≥10 years. We hypothesize that this observation is driven by survival bias. As mentioned earlier, the estimated average time to malignant transformation of OPMDs is 4–10 years [8, 9]. The paucity of HIV-specific data makes it difficult to surmise if average time would be significantly different in PLHIV. Compared to the general population, oral cancer is reported at younger ages in PLHIV [2], and it is plausible that the average time to malignant transformation is also reduced. Thus, PLHIV who have used SLT≥10 years, could be healthier than an average person living with HIV, given their longevity, contributing to the survival bias. Future longitudinal studies should seek to improve on our findings.

There are several limitations to our study findings. Due to the cross-sectional nature of our study design, risks of OPMDs for SLT use could not be calculated. Nevertheless, considering the extensive evidence that exists for the associations between OPMDs and SLT [6], SLT use and HIV [12], and oral cancer and HIV [2–4], our findings have high biological plausibility. Furthermore, by using PRs, we mitigated the overestimation of associations and potentially provide approximations of risk ratios. There were many challenges to study implementation in our setting, namely, low levels of agreement among clinicians and low follow-up compliance of participants, leading to misclassification of the outcome variable, and affecting estimations of prevalence and PRs. Structured clinician trainings to increase inter-clinician agreement and strategies to improve participant retention should be planned in future iterations of similar studies. Nonetheless, we addressed the issues of outcome misclassification by using validity parameters and analytical techniques described in literature [15–19] and generated a range of possible prevalence and PR estimates for our study population. The OPMD prevalence estimates for HIV-uninfected individuals were consistent with estimates reported elsewhere from India [14, 27–30], and the PRs generated were robust to the range of validity parameters used, strengthening our confidence in the results presented. We did not observe a significant association for former SLT users. While this could be related to the frequent resolution of OPMDs with the cessation of SLT use [12], it is possible that we are not adequately powered to detect this relationship. As we did not collect data on the frequency of SLT use, we were unable to assess if the associations observed for SLT use duration are modified by use frequency. Findings from the GATS-2 indicate that 85% of current SLT users in India are daily users and our results are unlikely to vary substantially by use frequency [11]. Smoked tobacco and alcohol use could have been underreported, notably for women, because of their lower social acceptability. We did not collect information on the socioeconomic conditions of participants and are unable to evaluate the extent to which they affect our associations. In India, low socioeconomic position is a determinant of both SLT use and oral cancer mortality [9, 26], and socioeconomic status should be accounted for in future analyses. Lastly, we did not further classify suspected OPMDs into different types, or SLT into different forms. We believe that these details are inconsequential to generate data to plan preventive measures, given that management strategies are similar across different OPMDs and SLT forms [31, 32].

Conclusion

SLT use is predominantly concentrated in LMICs, and its use is higher among PLHIV compared to HIV uninfected individuals in these countries [12]. However, as a public health problem, SLT use among PLHIV has received considerably less attention than smoking and alcohol use. Viewed through the perspective of health equity, our findings highlight the need to plan oral cancer screening programs for PLHIV in India with a priority for current SLT users, considering that such individuals possibly exist at the intersection of increased oral cancer risk, poverty, and HIV-related social vulnerability.

Supporting information

S1 File

(DOCX)

Click here for additional data file.^{(14KB, docx)}

S2 File

(DOCX)

Click here for additional data file.^{(3.3MB, docx)}

Acknowledgments

The authors would like to thank Dr. Radhika Chigurupati at Boston University and Dr. Robert Bollinger at Johns Hopkins University, co-developers of the m-Health application used in this study. We would like to specially acknowledge Dr. Radhika Chigurupati for all her help in the initial setting up of the m-Health platform. We thank Dr. Matthew Robinson for providing valuable insights on the first draft of this manuscript. We extend our gratitude to the team at Mueva el Volante especially Markus Aulkemeier for all the technical assistance; Suhasini Surwase and Archana Pawar, the health care workers on this study; Rohini Kamble for her assistance with data entry, and the entire staff at the ART centre. Most of all the authors would like to thank the participants on this study, without whom this work would not be possible.

Data Availability

Data cannot be shared publicly because it comes from people living with HIV in India, many of whom have not disclosed their HIV status except to their HIV health care provider and spouses. Data are available from the Ethics Committee - Byramjee Jeejeebhoy Medical College & Sassoon General Hospitals, Pune, India (contact via email @bjmcecirb@gmail.com) for researchers who meet the criteria for access to confidential data.

Funding Statement

This work was supported by amfAR, The Foundation for AIDS Research, with support from the U.S. National Institutes of Health’s National Institute of Allergy and Infectious Diseases, the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Cancer Institute, the National Institute of Mental Health, the National Institute on Drug Abuse, the National Heart, Lung, and Blood Institute, the National Institute on Alcohol Abuse and Alcoholism, the National Institute of Diabetes and Digestive and Kidney Diseases, and the Fogarty International Center, as part of the International Epidemiology Databases to Evaluate AIDS (IeDEA; U01AI069907) and the NIH-funded Johns Hopkins Baltimore-Washington-India Clinical Trials Unit for NIAID Networks (UM1AI069465). The content and views expressed are those of the authors and does not necessarily represent the official views of any of the governments or institutions mentioned above. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.GLOBOCAN, WHO, International Agency for Research on Cancer. Lip, oral cavity cancer incidence and mortality from 2020 to 2040. https://gco.iarc.fr/tomorrow/en/dataviz/isotype?cancers=1&single_unit=50000. Published 2021. Accessed April 3, 2021.
2.Shiels MS, Althoff KN, Pfeiffer RM, et al. HIV Infection, Immunosuppression, and Age at Diagnosis of Non-AIDS-Defining Cancers. Clin Infect Dis. 2017;64(4):468–475. doi: 10.1093/cid/ciw764 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Beachler DC, Abraham AG, Silverberg MJ, et al. Incidence and risk factors of HPV-related and HPV-unrelated Head and Neck Squamous Cell Carcinoma in HIV-infected individuals. Oral Oncol. 2014;50(12):1169–1176. doi: 10.1016/j.oraloncology.2014.09.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Godbole SV, Nandy K, Gauniyal M, et al. HIV and cancer registry linkage identifies a substantial burden of cancers in persons with HIV in India. Medicine (Baltimore). 2016;95(37):e4850. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Government of India MoHFW. Operational Framework: Management of Common Cancers. http://nhsrcindia.org/sites/default/files/Operational%20Framework%20Management%20of%20Common%20Cancers.pdf. Published 2016. Accessed March 31, 2021.
6.Khan Z, Khan S, Christianson L, Rehman S, Ekwunife O, Samkange-Zeeb F. Smokeless Tobacco and Oral Potentially Malignant Disorders in South Asia: A Systematic Review and Meta-analysis. Nicotine Tob Res. 2017;20(1):12–21. doi: 10.1093/ntr/ntw310 [DOI] [PubMed] [Google Scholar]
7.Rajaraman P, Anderson BO, Basu P, et al. Recommendations for screening and early detection of common cancers in India. The Lancet Oncology. 2015;16(7):e352–e361. doi: 10.1016/S1470-2045(15)00078-9 [DOI] [PubMed] [Google Scholar]
8.Mohan P, Richardson A, Potter JD, Coope P, Paterson M. Opportunistic Screening of Oral Potentially Malignant Disorders: A Public Health Need for India. JCO Glob Oncol. 2020;6:688–696. doi: 10.1200/JGO.19.00350 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Government of India MoHFW. Smokeless Tobacco and Public Health in India. https://untobaccocontrol.org/kh/smokeless-tobacco/smokeless-tobacco-public-health-india/. Published 2016. Accessed March 15, 2021.
10.Mello FW, Miguel AFP, Dutra KL, et al. Prevalence of oral potentially malignant disorders: A systematic review and meta-analysis. J Oral Pathol Med. 2018;47(7):633–640. doi: 10.1111/jop.12726 [DOI] [PubMed] [Google Scholar]
11.Government of India MoHFW, WHO, TISS. Global Adult Tobacco Survey: Second Round: India 2016–2017. https://ntcp.nhp.gov.in/assets/document/surveys-reports-publications/Global-Adult-Tobacco-Survey-Second-Round-India-2016-2017.pdf. Published 2018. Accessed April 1, 2021.
12.Mdege ND, Shah S, Ayo-Yusuf OA, Hakim J, Siddiqi K. Tobacco use among people living with HIV: analysis of data from Demographic and Health Surveys from 28 low-income and middle-income countries. The Lancet Global Health. 2017;5(6):e578–e592. doi: 10.1016/S2214-109X(17)30170-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.WHO Report on the Global Tobacco Epidemic, 2019. Geneva: World Health Organization; 2019. Licence: CC BY-NC-SA 3.0 IGO.
14.Birur PN, Sunny SP, Jena S, et al. Mobile health application for remote oral cancer surveillance. J Am Dent Assoc. 2015;146(12):886–894. doi: 10.1016/j.adaj.2015.05.020 [DOI] [PubMed] [Google Scholar]
15.Birur PN, Patrick S, Bajaj S, et al. A Novel Mobile Health Approach to Early Diagnosis of Oral Cancer. J Contemp Dent Pract. 2018;19(9):1122–1128. [PMC free article] [PubMed] [Google Scholar]
16.Alabdullah JH, Daniel SJ. A Systematic Review on the Validity of Teledentistry. Telemed J E Health. 2018;24(8):639–648. doi: 10.1089/tmj.2017.0132 [DOI] [PubMed] [Google Scholar]
17.Vinayagamoorthy K, Acharya S, Kumar M, Pentapati KC, Acharya S. Efficacy of a remote screening model for oral potentially malignant disorders using a free messaging application: A diagnostic test for accuracy study. Aust J Rural Health. 2019;27(2):170–176. doi: 10.1111/ajr.12496 [DOI] [PubMed] [Google Scholar]
18.Haron N, Zain RB, Nabillah WM, et al. Mobile Phone Imaging in Low Resource Settings for Early Detection of Oral Cancer and Concordance with Clinical Oral Examination. Telemed J E Health. 2017;23(3):192–199. doi: 10.1089/tmj.2016.0128 [DOI] [PubMed] [Google Scholar]
19.Lash T, Fox M, Fink A. Disease Misclassification, Corrections with Sensitivity and Specificity: Nondifferential and Independent Errors. In: Applying Quantitative Bias Analysis to Epidemiologic Data. Springer; 2009:94–96. [Google Scholar]
20.Lyles RH, Tang L, Superak HM, et al. Validation data-based adjustments for outcome misclassification in logistic regression: an illustration. Epidemiology. 2011;22(4):589–597. doi: 10.1097/EDE.0b013e3182117c85 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Shaw PA, Gustafson P, Carroll RJ, et al. STRATOS guidance document on measurement error and misclassification of variables in observational epidemiology: Part 2-More complex methods of adjustment and advanced topics. Stat Med. 2020;39(16):2232–2263. doi: 10.1002/sim.8531 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Keogh RH, Shaw PA, Gustafson P, Carroll RJ, Deffner V, Dodd KW, et al. STRATOS guidance document on measurement error and misclassification of variables in observational epidemiology: Part 1-Basic theory and simple methods of adjustment. Stat Med. 2020;39(16):2197–231. doi: 10.1002/sim.8532 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Liao LJ, Chou HL, Lo WC, et al. Initial outcomes of an integrated outpatient-based screening program for oral cancers. Oral Surg Oral Med Oral Pathol Oral Radiol. 2015;119(1):101–106. doi: 10.1016/j.oooo.2014.09.020 [DOI] [PubMed] [Google Scholar]
24.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. doi: 10.1016/j.oraloncology.2012.11.004 [DOI] [PubMed] [Google Scholar]
25.Cheung LC, Ramadas K, Muwonge R, et al. Risk-Based Selection of Individuals for Oral Cancer Screening. J Clin Oncol. 2021;39(6):663–674. doi: 10.1200/JCO.20.02855 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Mallath MK, Taylor DG, Badwe RA, et al. The growing burden of cancer in India: epidemiology and social context. The Lancet Oncology. 2014;15(6):e205–e212. doi: 10.1016/S1470-2045(14)70115-9 [DOI] [PubMed] [Google Scholar]
27.Kumar S, Debnath N, Ismail MB, Kumar A, Kumar A, Badiyani BK, et al. Prevalence and Risk Factors for Oral Potentially Malignant Disorders in Indian Population. Adv Prev Med. 2015;2015:20851 doi: 10.1155/2015/208519 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Sivakumar TT, Sam N, Joseph AP. Prevalence of oral potentially malignant disorders and oral malignant lesions: A population-based study in a municipal town of southern Kerala. J Oral Maxillofac Pathol. 2018;22(3):413–4. doi: 10.4103/jomfp.JOMFP_202_17 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Ramesh RM, Patrick S, Lotha Z, Azole Aier A, Birur NP. Prevalence and determinants of oral potentially malignant lesions using mobile health in a rural block, northeast India. Trop Doct. 2022;52(1):53–60. doi: 10.1177/00494755211049973 [DOI] [PubMed] [Google Scholar]
30.Misra V, Singh PA, Lal N, Agarwal P, Singh M. Changing pattern of oral cavity lesions and personal habits over a decade: hospital based record analysis from allahabad. Indian J Community Med. 2009;34(4):321–5. doi: 10.4103/0970-0218.58391 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Nadeau C, Kerr AR. Evaluation and Management of Oral Potentially Malignant Disorders. Dent Clin North Am. 2018;62(1):1–27. doi: 10.1016/j.cden.2017.08.001 [DOI] [PubMed] [Google Scholar]
32.Warnakulasuriya S. Oral potentially malignant disorders: A comprehensive review on clinical aspects and management. Oral Oncol. 2020;102:104550. doi: 10.1016/j.oraloncology.2019.104550 [DOI] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0270876.r001

Decision Letter 0

Jitendra Kumar Meena

24 Mar 2022

PONE-D-21-25906Smokeless tobacco use and oral potentially malignant disorders among people living with HIV (PLHIV) in Pune, India: Implications for oral cancer screening in PLHIVPLOS ONE

Dear Dr. Marbaniang,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by May 08 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Jitendra Kumar Meena

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. Please note that all PLOS journals ask authors to adhere to our policies for sharing of data and materials: https://journals.plos.org/plosone/s/data-availability. According to PLOS ONE’s Data Availability policy, we require that the minimal dataset underlying results reported in the submission must be made immediately and freely available at the time of publication. As such, please remove any instances of 'unpublished data' or 'data not shown' in your manuscript and replace these with either the relevant data (in the form of additional figures, tables or descriptive text, as appropriate), a citation to where the data can be found, or remove altogether any statements supported by data not presented in the manuscript.

3. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match. When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

4. Thank you for stating the following in the Funding Section of your manuscript:

“This work was supported by amfAR, The Foundation for AIDS Research, with support from the U.S. National Institutes of Health’s National Institute of Allergy and Infectious Diseases, the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Cancer Institute, the National Institute of Mental Health, the National Institute on Drug Abuse, the National Heart, Lung, and Blood Institute, the National Institute on Alcohol Abuse and Alcoholism, the National Institute of Diabetes and Digestive and Kidney Diseases, and the Fogarty International Center, as part of the International Epidemiology Databases to Evaluate AIDS (IeDEA; U01AI069907 as a sub-grant to VM) and the NIH-funded Johns Hopkins Baltimore-Washington-India Clinical Trials Unit for NIAID Networks [UM1AI069465 to VM]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“This work was supported by amfAR, The Foundation for AIDS Research, with support from the U.S. National Institutes of Health’s National Institute of Allergy and Infectious Diseases, the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Cancer Institute, the National Institute of Mental Health, the National Institute on Drug Abuse, the National Heart, Lung, and Blood Institute, the National Institute on Alcohol Abuse and Alcoholism, the National Institute of Diabetes and Digestive and Kidney Diseases, and the Fogarty International Center, as part of the International Epidemiology Databases to Evaluate AIDS (IeDEA; U01AI069907 as a sub-grant to VM) and the NIH-funded Johns Hopkins Baltimore-Washington-India Clinical Trials Unit for NIAID Networks [UM1AI069465]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

5. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

6. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This is a very well-written and thoughtful manuscript. Kudos to the authors for a valuable contribution to the literature. A few suggestions and questions below:

Line 246-250: the low f/u rate (46%), combined with “poor to slight” agreement among clinicians on the OPMD assessment presents a potential challenge in drawing inferences about the outcome variable and its relationship to SLT, HIV status, etc. Perhaps the authors could better explain how the use of validity parameters mitigated this challenge.

Line 173-179: although PRs for OPMD are reported as higher among PLHIV in both adjusted and unadjusted models, the prevalence estimate ranges were reported as overlapping (16.3-48.7 for PLHIV, and 6.8-18.5 for HIV uninfected) when corrected. Estimate range was also very broad for PLHIV in particular. Does this have an impact upon the strength of the conclusions drawn? Are PRs for PLHIV vs. HIV-uninfected not affected by the increased SLT duration noted among the former? If so, this should be mentioned in the Discussion section.

Lines 196-200: while OPV and gender are mentioned, there is no mention of CD4 count. Intuitively, wouldn’t one assume that lower CD4 count was associated with higher rates of OPMD? Perhaps there should be some mention of whether or not this was the case.

Line 268: poverty is mentioned here for the first time (except that BJGMC-SGH caters to low and lower-middle SES patients, line 63-64). This should be mentioned among limitations, as it’s possible that SLT use and/or HIV status in this area is associated with lower SES and/or lower health service utilization, representing a potential confounding contributor to OPMD/oral cancer risk.

Reviewer #2: SLT use among PLHIV is an important public health problem given it’s potential impact on morbidity and mortality of PLHIV. This study aims to “compare the prevalence of OPMDs among PLHIV to HIV uninfected individuals; quantify the association of OPMDs with SLT use among PLHIV; and estimate the association between OPMDs and SLT use for current Users..”

Below are comments and suggested edits

43 Due to its cultural acceptability across genders and high prevalence of use, smokeless

44 tobacco (SLT) is a predominant cause of OPMDs in India [6,9].

Please change consider changing “cause” to risk factor

45………………………………………….. ………………………………..For PLHIV, results

46 from a meta-analysis indicate that the prevalence of SLT use could be twice as high as the general

47 population [11].

Is this just in India or globally? Please clarify (I don’t believe this is true among most PLHIV globally)

It is unclear how the sample size for this study was calculated; what assumptions were made about prevalence of SLT use among PLHIV?

167 to make a diagnosis. The kappa statistic ranged between 0.16 to 0.33, indicating poor to slight

168 agreement among clinicians.

This seems like a problem that casts a big doubt on the main methodology used in this study to assess outcomes of interest including determining the prevalence of suspected OPMDs and link between SLT use and suspected OPMDs among PLHIV. How do you explain the huge discrepancies in results from reading images among the clinicians? In one of the provided references (Birur PN, Sunny SP, Jena S, et al.), the concordance of results between the specialists was almost 100%. I think it is critical to explain the reasons for the poor agreement and how it may have impacted the results.

170 The prevalence of suspected OPMDs was 15% (n=186) for the entire study population,

171 19% (n=117) for PLHIV and 11% (n=69) for the HIV uninfected group.

Again, it is hard to interpret these results without knowing some of the assumptions that went into sample size calculations. For example, what assumptions were made about the agreement between the clinicians in reading images.

176 In the unadjusted model, relative to HIV uninfected individuals, the prevalence of

177 suspected OPMDs among PLHIV was 1.79 (95% CI: 1.36 – 2.35) times higher. When adjusted for covariates, the association remained statistically significant,

In the discussion section the authors did not address the potential impact of HPV on these estimates even though globally HPV prevalence is known to be higher among PLHIV and the association between HPV and cancers (including oral cancers) is well documented.

219 A similar approach culturally adapted for India and integrated into routine HIV care could greatly

220 benefit PLHIV that are current SLT users.

Given the relatively high prevalence of suspected OPMDs among all SLT users (yes higher rates among PLHIV compared with un-infected), wouldn’t it make sense to make this recommendation for all STL users. Yes it is important to prioritize but there are relatively fewer PLHIV SLT users in India compared to the general public who use SLT and it would seem to me that broader public health work around SLT prevention will have a bigger impact compared to limiting work on PLHIV who use STL. This may reduce the risk of coming up with fragmented strategies to address this problem among the different “high risk groups”.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Shirish Balachandra

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2022 Jul 5;17(7):e0270876. doi: 10.1371/journal.pone.0270876.r002

Author response to Decision Letter 0

2 Jun 2022

We thank the reviewers for their comments. The comments raised were thoughtful, and they greatly helped us in improving the shortcomings of the previous version of this manuscript. We have tried to address the reviewers’ comments to the best of our ability. If we have been unable to address them, we have added these as limitations of our work. Since the subject of smokeless tobacco (SLT) use and oral potentially malignant disorders (OPMDs) among people living with HIV (PLHIV) has not been very well explored, we hope that this revised version of the manuscript will be considered favorably by the reviewers.

Please note that the line numbers presented for the reviewers are as they appeared in the original manuscript that had been submitted. Wherever the line numbers are as they appear in the revised manuscript without track changes, this has been explicitly mentioned.

Reviewer 1

Comment 1: Line 246-250: the low f/u rate (46%), combined with “poor to slight” agreement among clinicians on the OPMD assessment presents a potential challenge in drawing inferences about the outcome variable and its relationship to SLT, HIV status, etc. Perhaps the authors could better explain how the use of validity parameters mitigated this challenge.

Authors’ response: We thank the reviewer for this comment. We agree that the in the originally submitted manuscript the utility of including validity parameters was not explained at length. We have tried our best to clarify how the inclusion of validity parameters was utilized in the analysis and how this is helpful (lines 202 – 229 of the revised manuscript without track changes).

Briefly, we had two major issues when trying to draw inferences about the associations between HIV status, SLT use and OPMDs. These were:

a) A low level of agreement between reviewing clinicians, as indicated by the low kappa statistic.

b) High non-compliance with follow-up procedures for in-person clinical examination, as indicated by the low follow-up.

Both these issues produced misclassification of the outcome variable, albeit in different ways. For example: a) produced misclassification of suspected OPMDs - i.e., participants that would have been classified to have suspected OPMDs were not and vice versa, b) produced misclassification of the outcome variable in relation to in-person clinical examination, i.e., participants that would have been deemed to be negative for OPMDs on in-person clinical examination were classified as OPMD positive and vice versa. Of these two issues, we believe that the non-compliance with follow-up procedures posed a more pressing problem, given that even if the kappa statistic had been 1 (perfect agreement between clinicians), estimates for prevalence and prevalence ratios (PRs) would have been based on images review i.e., suspected OPMDs, whereas to inform clinical practice and policy, we would need to understand how the suspected OPMDs compared against in-person clinical examination diagnoses.

Therefore, we viewed the distribution of the outcome variable obtained under low inter-clinician agreement as one arising from a dataset where there was misclassification of the outcome variable relative to clinical examination. To correct for this misclassification, we used an external set of sensitivity and specificity parameters (derived from studies that had compared images review findings to in-person clinical examination) and obtained a plausible range of estimates for prevalence and PRs. The methodology on how this was done has been described in Part A of the supplementary file.

Using these validity parameters allowed us to provide corrected prevalence estimates and corrected adjusted PRs that may be more useful to draw out clinical inferences for our study population.

Comment 2: Line 173-179: although PRs for OPMD are reported as higher among PLHIV in both adjusted and unadjusted models, the prevalence estimate ranges were reported as overlapping (16.3-48.7 for PLHIV, and 6.8-18.5 for HIV uninfected) when corrected. Estimate range was also very broad for PLHIV in particular. Does this have an impact upon the strength of the conclusions drawn? Are PRs for PLHIV vs. HIV-uninfected not affected by the increased SLT duration noted among the former? If so, this should be mentioned in the Discussion section.

Authors’ response: We thank the reviewer for this valuable comment. We recognize that in the originally submitted manuscript it was unclear what the range of the prevalence estimates indicated. We have tried to make this clearer by including Fig 2 in the revised manuscript and shown select numerical prevalence estimates at certain specificity and sensitivity parameters. The range stated for the prevalence estimates is not interpreted as 95% Confidence Intervals would be, thus they cannot be said to overlap, i.e., they do not represent the range of possible prevalence estimates accounting for some level of uncertainty. These are estimates of the prevalence at very specific sensitivity and specificity parameters.

For example: For PLHIV, a prevalence of 16.3% would only be observed if the sensitivity of the test (comparing diagnosis of OPMD by images review versus in-person clinical examination) is 0.99 and the specificity of the test is 0.96. At these sensitivity and specificity parameters, the only possible prevalence of OPMDs for HIV-uninfected individuals is 6.8%.

Similarly, for PLHIV, a prevalence of 48.7% is possible only if the sensitivity of the test is 0.39 and the specificity 0.99. At these sensitivity and specificity parameters, the only possible prevalence of OPMDs for HIV-uninfected individuals is 24.9% (this figure has been corrected from the previous figure of 18.5%). If we turn our attention to the Tables below (which are included as part of Figure 2), we will observe that for a given set of sensitivity and specificity parameters, the corrected prevalence estimates between PLHIV and HIV-uninfected individuals do not overlap.

Prevalence estimates for PLHIV (%) at specific sensitivity and specificity parameters

Sp=0.96, Se=0.39 Sp=0.97, Se=0.39 Sp=0.98, Se=0.39 Sp=0.99, Se=0.39

Prevalence= 44.3 Prevalence= 45.8 Prevalence= 47.3 Prevalence= 48.7

Sp=0.96, Se=0.54 Sp=0.97, Se=0.54 Sp=0.98, Se=0.54 Sp=0.99, Se=0.54

Prevalence= 30.9 Prevalence= 0.32 Prevalence= 33.6 Prevalence= 35.0

Sp=0.96, Se=0.69 Sp=0.97, Se=0.69 Sp=0.98, Se=0.69 Sp=0.99, Se=0.69

Prevalence= 23.8 Prevalence= 0.25 Prevalence= 26.1 Prevalence= 27.2

Sp=0.96, Se=0.84 Sp=0.97, Se=0.84 Sp=0.98, Se=0.84 Sp=0.99, Se=0.84

Prevalence= 19.4 Prevalence= 20.4 Prevalence= 21.3 Prevalence= 22.2

Sp=0.96, Se=0.99 Sp=0.97, Se=0.99 Sp=0.98, Se=0.99 Sp=0.99, Se=0.99

Prevalence= 16.3 Prevalence= 17.2 Prevalence= 18.0 Prevalence= 18.9

Prevalence estimates for HIV-uninfected (%) at specific sensitivity and specificity parameters

Sp=0.96, Se=0.39 Sp=0.97, Se=0.39 Sp=0.98, Se=0.39 Sp=0.99, Se=0.39

Prevalence= 18.5 Prevalence= 20.8 Prevalence= 22.9 Prevalence= 24.9

Sp=0.96, Se=0.54 Sp=0.97, Se=0.54 Sp=0.98, Se=0.54 Sp=0.99, Se=0.54

Prevalence= 12.9 Prevalence= 14.7 Prevalence= 16.3 Prevalence= 17.9

Sp=0.96, Se=0.69 Sp=0.97, Se=0.69 Sp=0.98, Se=0.69 Sp=0.99, Se=0.69

Prevalence= 9.9 Prevalence= 11.3 Prevalence= 12.7 Prevalence= 13.9

Sp=0.96, Se=0.84 Sp=0.97, Se=0.84 Sp=0.98, Se=0.84 Sp=0.99, Se=0.84

Prevalence= 8.1 Prevalence= 9.2 Prevalence= 10.4 Prevalence= 11.4

Sp=0.96, Se=0.99 Sp=0.97, Se=0.99 Sp=0.98, Se=0.99 Sp=0.99, Se=0.99

Prevalence= 6.8 Prevalence= 7.8 Prevalence= 8.7 Prevalence= 9.7

We do not believe that the different prevalence estimates by themselves (depending on the validity parameters) affect the strength of our conclusions drawn, as the corrected adjusted PRs comparing PLHIV and HIV-uninfected individuals remains fairly constant (between 1.38 – 1.64 under different validity parameters, lines 276 - 278 of the revised manuscript without track changes).

These findings indicate the importance in recognizing that PLHIV are potentially at higher risk of OPMDs than HIV-uninfected individuals, and the need to consider oral cancer screening integration into HIV care in India. Where we believe different prevalence estimates could play a role is in resource allocation. To understand this better, larger studies/surveys where the prevalence of OPMDs among PLHIV are ascertained more rigorously are required.

Lastly, we addressed the association between SLT use, SLT duration of use, HIV status and OPMDs by stratifying on duration of SLT use (<10 years, ≥10 years) and HIV status in our analysis. These results are presented in Fig 4. As we observe in Fig 4 and have discussed (lines 330 – 340 of the revised manuscript without track changes), we observe a significant positive association (PR: 3.46, 95% CI: 1.48, 8.09) between HIV status and OPMDs for current SLT users when the duration of use is <10 years, but not for when use duration is ≥10 years (PR: 1.28, 95% CI: 0.90, 1.81). If the association between OPMDs and HIV seropositivity was entirely affected by the duration of SLT use, then we would have seen a consistent significant positive relationship even for duration of use ≥10 years. As we discuss (lines 333-340 of the revised manuscript without track changes) we believe the non-significant relationship observed for ≥10 years duration of use is driven partly by selection bias (i.e., this being a cross-sectional analysis, we have selected PLHIV who may be healthier than PLHIV on average, as they have been able to live with HIV, but given their survival have also been able to use SLT longer). We would also like to mention however that even if we had been able to account for selection bias in our analysis, the association with duration of SLT use is a complex issue to discuss. Currently there are no standard methods to assess SLT use duration while also accounting for the quantity of SLT used (like pack-years for cigarette smoking). SLT use duration by itself, we believe is a broad and imperfect measure. Therefore, we have refrained from making too many inferences in this manuscript with respect to the duration of SLT use.

Comment 3: Lines 196-200: while OPV and gender are mentioned, there is no mention of CD4 count. Intuitively, wouldn’t one assume that lower CD4 count was associated with higher rates of OPMD? Perhaps there should be some mention of whether or not this was the case.

Authors’ response: Thank you for this comment. The objectives of this manuscript (as stated in lines 89 to 92 of the revised manuscript without track changes) are primarily to quantify the associations between SLT use and OPMDs in PLHIV. We had originally mentioned oral HPV and gender in the lines that the reviewer cites as they relate to findings of the sensitivity analyses conducted i.e., how including them as covariates in multivariable models affects the primary associations of interest. We have made it clearer in the revised manuscript (without track changes, lines 298 – 302) that findings related to these two covariates are as they pertain to the sensitivity analyses i.e., their inclusion does not affect the primary associations of interest (estimates are presented in S2 Table of the supplementary files). We agree with the reviewer that CD4 count is an important factor to consider in relation to OPMDs, that is why it is included as a covariate in the multivariable model when the analysis is restricted to PLHIV (Table 2, Model 2). In our analysis we did not find significant associations between CD4 count and OPMDs either in the univariate or multivariable models (Table 2, Model 2) among PLHIV. As CD4 count is not conventionally measured in HIV-uninfected individuals, it was not included as a covariate in models where comparisons with HIV-uninfected individuals were made.

Comment 4: Line 268: poverty is mentioned here for the first time (except that BJGMC-SGH caters to low and lower-middle SES patients, line 63-64). This should be mentioned among limitations, as it’s possible that SLT use and/or HIV status in this area is associated with lower SES and/or lower health service utilization, representing a potential confounding contributor to OPMD/oral cancer risk.

Authors’ response: We thank the reviewer for this important comment. We have included the lack of socioeconomic information as a limitation in the revised manuscript (lines 363 – 366 of the revised manuscript without track changes).

Reviewer 2

Comment 5: Line 43 – 44 Due to its cultural acceptability across genders and high prevalence of use, smokeless tobacco (SLT) is a predominant cause of OPMDs in India [6,9]. Please change consider changing “cause” to risk factor.

Authors’ response: We thank the reviewer for this comment. This modification has been made (line 75-76 of the revised manuscript without track changes).

Comment 6: Line 45 – 47: For PLHIV, results from a meta-analysis indicate that the prevalence of SLT use could be twice as high as the general population.

Is this just in India or globally? Please clarify (I don’t believe this is true among most PLHIV globally)

Authors’ response: Thank you for this comment. We have further qualified the data used in the meta-analysis and corrected the estimate to mention that the prevalence of SLT use could be 1.3 times higher among PLHIV than the general population (line 77 – 80 of the revised manuscript without track changes).

Briefly, the prevalence of SLT in this meta-analysis was estimated using data from the Demographic and Health Survey (DHS) which included 18,224 individuals residing in 28 low- and middle-income countries (LMICs). Authors found that the relative risk (RR) of SLT use comparing PLHIV to HIV-uninfected individuals was 1·32 [95% CI: 1·03–1·69]; p=0·030 among women and (1·26 [95% CI:1·00–1·58]; p=0·050) among men.

Comment 7: It is unclear how the sample size for this study was calculated; what assumptions were made about prevalence of SLT use among PLHIV?

Authors’ response: We would like to thank the reviewer for this important comment. It has allowed us to clarify the assumptions made for the sample size of the study population. These have been mentioned in lines 104 -115 of the revised manuscript without track changes.

• To calculate the sample sizes for PLHIV and HIV-uninfected participants to be enrolled, we first made two assumptions:

a) The prevalence of OPMDs among HIV-uninfected individuals was assumed to be 10.5%, as reported in a meta-analysis for Asian populations by Mello FW (J Oral Pathol Med. 2018;47(7):633-640)

b) The prevalence ratio for OPMDs was assumed to be 1.5 among PLHIV compared to HIV-uninfected individuals.

• We then fixed the two-sided confidence interval at 95%, and an enrolment ratio between HIV-uninfected individuals and PLHIV to 1:1. Under these parameters we expected to enrol 1320 participants overall (660 PLHIV and 660 HIV-uninfected participants) and be 80% powered to detect an OPMD prevalence of 15.7% among PLHIV.

• However, as the study proceeded, we had budgetary constraints and we were unable to meet the original enrolment targets. We therefore revised the power to 77% and the required sample size to 1226 (613 PLHIV and 613 HIV-uninfected participants).

• In the end, we were able to enrol 601 PLHIV and 633 HIV-uninfected individuals. This changed the enrolment ratio to 1.05:1. The other parameters remaining the same, this allowed us to observe an OPMD prevalence of 15.7% among PLHIV powered at 77%.

• For the study population accrued (i.e., 601 PLHIV and 633 HIV-uninfected individuals), assuming the prevalence of SLT use in the general population to be 21.4% (2016-2017 Global Adult Tobacco Survey estimate, line 74 of the revised manuscript without track changes) and the prevalence ratio of SLT use to be 1.32 times higher among PLHIV compared to HIV-uninfected individuals (The Lancet Global Health. 2017;5(6):e578-e592 estimates, line 80 of the revised manuscript without track changes), the study is 80% powered to detect a prevalence of SLT use of 28.4% among PLHIV.

Comment 8: Line 167 – 168: to make a diagnosis. The kappa statistic ranged between 0.16 to 0.33, indicating poor to slight agreement among clinicians.

Authors’ response: We thank the reviewer for this comment. We acknowledge that the low kappa statistic posed challenges in determining the prevalence of OPMDs in this study population. However, we believe that the sensitivity analyses we conducted by using validity parameters and the methodology proposed by Lash et al. (Lash T, Fox M, Fink A. Disease Misclassification, Corrections with Sensitivity and Specificity: Nondifferential and Independent Errors. In: Applying Quantitative Bias Analysis to Epidemiologic Data. Springer; 2009:94-96) helped us to provide a plausible range of corrected prevalence estimates for OPMDs.

The range of corrected prevalence estimates in the HIV-uninfected population (between 6.8% - 24.9%, Fig 2) obtained using the methodology described by Lash et al. corresponds to estimates that have been reported from the country (shown below), which increases our confidence in the results presented.

Study sample size Prevalence of OPMDs

Birur, Praveen N et al. “Mobile health application for remote oral cancer surveillance.” Journal of the American Dental Association (1939) vol. 146,12 (2015): 886-94 1440 7.4%

Kumar S, Debnath N, Ismail MB, et al. Prevalence and Risk Factors for Oral Potentially Malignant Disorders in Indian Population. Adv Prev Med. 2015;2015:208519. 1241 13.7%

Sivakumar, T T et al. “Prevalence of oral potentially malignant disorders and oral malignant lesions: A population-based study in a municipal town of southern Kerala.” Journal of oral and maxillofacial pathology : JOMFP vol. 22,3 (2018): 413-414. 2368 6.6%

Ramesh, Rohan Michael et al. “Prevalence and determinants of oral potentially malignant lesions using mobile health in a rural block, northeast India.” Tropical doctor vol. 52,1 (2022): 53-60. 2686 26%

Misra, Vatsala et al. “Changing pattern of oral cavity lesions and personal habits over a decade: hospital based record analysis from allahabad.” Indian journal of community medicine : official publication of Indian Association of Preventive & Social Medicine vol. 34,4 (2009): 321-5 753 19.4%

We also acknowledge that while planning the study, we did not foresee the significant disagreement between clinicians. We assumed that given the significant number of years of clinical experience of the clinicians involved in managing Head and Neck conditions, there would have been high agreement, and therefore structured trainings were not warranted. The absence of these trainings, we believe could have significantly contributed to the low kappa statistic. We have included a line about there being no structured trainings for clinicians (lines 160 – 161 in the revised manuscript without track changes) and mentioned the absence of such training as a limitation which must be addressed if similar studies are planned in the future (lines 349 - 350).

In the Birur PN et at. article the concordance of 100% was calculated differently to how a kappa statistic is calculated (which accounts for the probability that agreement was reached by chance). In the article, only images that were deemed ‘positive’ by a single dental professional were further sent to specialists to be reviewed, and the concordance was based only on positive images. The reviewers were not shown the images of the participants that the single dental professional had deemed negative. In our study, two clinicians simultaneously reviewed the images, and in the event that the first two clinicians disagreed, a third clinician’s opinion was sought. If we calculate the concordance like they did in the Birur PN et.al paper, then we have a concordance of 87%.

Lastly, as we explained in our response to comment 1, there were two main issues when trying to draw inferences about the associations (PRs) between HIV status, SLT use and OPMDs. These were:

a) A high level of disagreement between reviewing clinicians as indicated by the low kappa statistic.

b) High non-compliance with follow-up procedures for in-person clinical examination.

Therefore, we viewed the distribution of the outcome variable obtained under low inter-clinician agreement as one arising from a dataset where there was misclassification of the outcome variable relative to clinical examination. To correct for this misclassification, we used a wide range of sensitivity and specificity parameters (derived from studies that had compared images review findings to in-person clinical examination) and obtained a plausible range of estimates for prevalence and PRs. The methodology on how this was done has been described in Part A of the supplementary file.

It would have been easier for us to state that the misclassification of the outcome variable (under non-differential misclassification) would bias the PRs towards the null (i.e., make the associations more conservative), but the sensitivity analyses that we performed allowed us to explicitly quantify the range of PRs plausible under a broad range of validity parameters for the study population.

We have expanded on how the low levels of agreement and low follow-up proportions could affect the estimates and how these were handled in the sensitivity analyses (lines 202 - 235 of the revised manuscript without track changes).

Comment 9: Line 170 – 171: The prevalence of suspected OPMDs was 15% (n=186) for the entire study population, 19% (n=117) for PLHIV and 11% (n=69) for the HIV uninfected group. Again, it is hard to interpret these results without knowing some of the assumptions that went into sample size calculations. For example, what assumptions were made about the agreement between the clinicians in reading images.

Authors’ response: Thank you for this comment. We have mentioned the assumptions that went into the sample size calculations in our response for comment 7. As we mention in comment 8, all clinicians involved in this study are well experienced in managing Head and Neck conditions and we assumed wrongly that there would be high agreement between (at least) two clinicians. Therefore, we did not explicitly account for clinician agreement when we were calculating the sample size requirements. If we had not been constrained by study costs, we would have increased the sample size to account for some part of the misclassification error during the period the study was actively enrolling. However, costs for processing and storing oral HPV samples steeply increased during the financial years of the study, beyond what had been budgeted for, which did not permit us from enrolling more participants.

Comment 10: Line 176 – 177: In the unadjusted model, relative to HIV uninfected individuals, the prevalence of suspected OPMDs among PLHIV was 1.79 (95% CI: 1.36 – 2.35) times higher. When adjusted for covariates, the association remained statistically significant.

Authors’ response: We thank the reviewer for this comment.

• The objectives of this manuscript (as stated in lines 89 to 92 of the revised manuscript without track changes) are primarily to quantify the associations between SLT use and OPMDs in PLHIV. Therefore, to be consistent with these objectives we do not believe that we necessarily need to discuss the association between oral HPV and OPMDs among PLHIV separately. However, we clearly present the association between oral HPV and OPMDs among PLHIV in Table 2 (Model 2). In our analysis, we found that oral HPV is not associated with suspected OPMDs among PLHIV in univariate analysis [Prevalence Ratio 0.66 (95% CI: 0.31 – 1.42)].

• As the reviewer has pointed out, we acknowledge that oral HPV is more prevalent among PLHIV and is associated with oral cancer. However, there were 233 and 19 missing oral HPV values among HIV-uninfected individuals and PLHIV, respectively, in our data. Therefore, as part of our sensitivity analyses, recognizing that oral HPV is an important covariate, we first imputed for the missing values before including it in a multivariable model (lines 235-239 of the revised manuscript without track changes). The results of the multivariable model in which oral HPV is included as a covariate are presented in S2 Table (supplementary files). We explicitly mention that the addition of the imputed values of oral HPV as a covariate did not affect our primary findings in the results section of our manuscript (lines 298 – 299 of the revised manuscript without track changes).

Comment 11: Line 219 - 220 A similar approach culturally adapted for India and integrated into routine HIV care could greatly benefit PLHIV that are current SLT users.

Given the relatively high prevalence of suspected OPMDs among all SLT users (yes higher rates among PLHIV compared with un-infected), wouldn’t it make sense to make this recommendation for all STL users. Yes, it is important to prioritize but there are relatively fewer PLHIV SLT users in India compared to the general public who use SLT and it would seem to me that broader public health work around SLT prevention will have a bigger impact compared to limiting work on PLHIV who use STL. This may reduce the risk of coming up with fragmented strategies to address this problem among the different “high risk groups”.

Authors’ response: We thank the reviewer for this insightful comment. We agree that in an ideal situation in which health services are equitably distributed, recommending an overarching and broad policy to screen all SLT users for OPMDs would equally benefit PLHIV. However, in the current Indian health care context, it is unlikely that recommending such a broad policy (i.e., the top – down approach) would benefit PLHIV, who are potentially at higher risk of OPMDs and oral cancer.

There are several factors related to HIV care in India, which we believe would not work in favor of a broad policy recommendation. Below we briefly mention two of these factors.

a. Logistical factors: India has the third highest number of PLHIV globally (estimated at 2.3 million). HIV related care is mostly delivered under programmatic conditions (treatment is subsidized by the government) in antiretroviral treatment (ART) centers, which are separate from outpatient clinics used by the general population. HIV medications are provided every 2 weeks to 3 months; CD4 counts/HIV viral RNA tests are performed every 6 months at the ART centers. The integration with other services – for example cancer screening programs is rare in the current HIV care delivery system. At present, PLHIV who wish to avail services like cancer screening need to book another appointment with a separate healthcare provider at a separate clinic. We recommend integrating oral cancer screening programs into routine HIV care in India because it would make access to these services easier (i.e., at the ART center, instead of PLHIV having to visit another health care provider) and can be done regularly (i.e., when PLHIV come for their medications or tests).

b. HIV-related discrimination in health care settings: Studies done across India indicate that HIV-related discrimination in health care settings persist in the country. A few examples are provided below.

• In a study that assessed cervical cancer screening among women living with HIV (WLHIV) in India, it was found that some clinicians involuntarily disclosed the HIV status of the women to others, while other health care providers prioritized HIV-uninfected women or actively discriminated against WLHIV. It was also observed that WLHIV chose to forgo recommended cervical cancer screening and treatment of cervical abnormalities because they were concerned about being discriminated against. Women Health. 2019;59(7):801-814

• In a qualitative study, authors highlight that health care providers often refused to treat PLHIV or referred them elsewhere. BMJ Open. 2019;9(11):e033790

• In a systematic review of 37 papers, authors highlighted discriminatory practices that persist in health care settings throughout India including minimising contact with PLHIV, denying PLHIV assistance during pregnancy, delaying treatment or care, and refusing treatment to PLHIV. SAHARA J. 2011;8(3):138-149

Please note that we are not arguing against oral cancer screening for SLT users in the general population. In fact, as we mention in the introduction (lines 59 - 60 of the revised manuscript without track changes), the government of India has formulated an operational framework for oral cancer management in which it proposes screening all adults>30 years for OPMDs every five years. However, given the persistence of HIV-related discrimination in health care settings in India and considering how HIV care is delivered (point (a) above), it is unlikely that PLHIV will avail oral cancer screening programs intended for the general population. Previous preventive strategies like tuberculosis preventive therapy (TPT) have been successfully integrated into HIV care (while still being available for other tuberculosis high risk groups). We recommend that a similar approach be adopted for oral cancer.

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(50.2KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0270876.r003

Decision Letter 1

Jitendra Kumar Meena

20 Jun 2022

Smokeless tobacco use and oral potentially malignant disorders among people living with HIV (PLHIV) in Pune, India: Implications for oral cancer screening in PLHIV

PONE-D-21-25906R1

Dear Dr. Marbaniang,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Jitendra Kumar Meena

Academic Editor

PLOS ONE

PLoS One. doi: 10.1371/journal.pone.0270876.r004

Acceptance letter

Jitendra Kumar Meena

24 Jun 2022

PONE-D-21-25906R1

Smokeless tobacco use and oral potentially malignant disorders among people living with HIV (PLHIV) in Pune, India: Implications for oral cancer screening in PLHIV

Dear Dr. Marbaniang:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Jitendra Kumar Meena

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 File

(DOCX)

Click here for additional data file.^{(14KB, docx)}

S2 File

(DOCX)

Click here for additional data file.^{(3.3MB, docx)}

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(50.2KB, docx)}

Data Availability Statement

[pone.0270876.ref001] 1.GLOBOCAN, WHO, International Agency for Research on Cancer. Lip, oral cavity cancer incidence and mortality from 2020 to 2040. https://gco.iarc.fr/tomorrow/en/dataviz/isotype?cancers=1&single_unit=50000. Published 2021. Accessed April 3, 2021.

[pone.0270876.ref002] 2.Shiels MS, Althoff KN, Pfeiffer RM, et al. HIV Infection, Immunosuppression, and Age at Diagnosis of Non-AIDS-Defining Cancers. Clin Infect Dis. 2017;64(4):468–475. doi: 10.1093/cid/ciw764 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref003] 3.Beachler DC, Abraham AG, Silverberg MJ, et al. Incidence and risk factors of HPV-related and HPV-unrelated Head and Neck Squamous Cell Carcinoma in HIV-infected individuals. Oral Oncol. 2014;50(12):1169–1176. doi: 10.1016/j.oraloncology.2014.09.011 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref004] 4.Godbole SV, Nandy K, Gauniyal M, et al. HIV and cancer registry linkage identifies a substantial burden of cancers in persons with HIV in India. Medicine (Baltimore). 2016;95(37):e4850. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref005] 5.Government of India MoHFW. Operational Framework: Management of Common Cancers. http://nhsrcindia.org/sites/default/files/Operational%20Framework%20Management%20of%20Common%20Cancers.pdf. Published 2016. Accessed March 31, 2021.

[pone.0270876.ref006] 6.Khan Z, Khan S, Christianson L, Rehman S, Ekwunife O, Samkange-Zeeb F. Smokeless Tobacco and Oral Potentially Malignant Disorders in South Asia: A Systematic Review and Meta-analysis. Nicotine Tob Res. 2017;20(1):12–21. doi: 10.1093/ntr/ntw310 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref007] 7.Rajaraman P, Anderson BO, Basu P, et al. Recommendations for screening and early detection of common cancers in India. The Lancet Oncology. 2015;16(7):e352–e361. doi: 10.1016/S1470-2045(15)00078-9 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref008] 8.Mohan P, Richardson A, Potter JD, Coope P, Paterson M. Opportunistic Screening of Oral Potentially Malignant Disorders: A Public Health Need for India. JCO Glob Oncol. 2020;6:688–696. doi: 10.1200/JGO.19.00350 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref009] 9.Government of India MoHFW. Smokeless Tobacco and Public Health in India. https://untobaccocontrol.org/kh/smokeless-tobacco/smokeless-tobacco-public-health-india/. Published 2016. Accessed March 15, 2021.

[pone.0270876.ref010] 10.Mello FW, Miguel AFP, Dutra KL, et al. Prevalence of oral potentially malignant disorders: A systematic review and meta-analysis. J Oral Pathol Med. 2018;47(7):633–640. doi: 10.1111/jop.12726 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref011] 11.Government of India MoHFW, WHO, TISS. Global Adult Tobacco Survey: Second Round: India 2016–2017. https://ntcp.nhp.gov.in/assets/document/surveys-reports-publications/Global-Adult-Tobacco-Survey-Second-Round-India-2016-2017.pdf. Published 2018. Accessed April 1, 2021.

[pone.0270876.ref012] 12.Mdege ND, Shah S, Ayo-Yusuf OA, Hakim J, Siddiqi K. Tobacco use among people living with HIV: analysis of data from Demographic and Health Surveys from 28 low-income and middle-income countries. The Lancet Global Health. 2017;5(6):e578–e592. doi: 10.1016/S2214-109X(17)30170-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref013] 13.WHO Report on the Global Tobacco Epidemic, 2019. Geneva: World Health Organization; 2019. Licence: CC BY-NC-SA 3.0 IGO.

[pone.0270876.ref014] 14.Birur PN, Sunny SP, Jena S, et al. Mobile health application for remote oral cancer surveillance. J Am Dent Assoc. 2015;146(12):886–894. doi: 10.1016/j.adaj.2015.05.020 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref015] 15.Birur PN, Patrick S, Bajaj S, et al. A Novel Mobile Health Approach to Early Diagnosis of Oral Cancer. J Contemp Dent Pract. 2018;19(9):1122–1128. [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref016] 16.Alabdullah JH, Daniel SJ. A Systematic Review on the Validity of Teledentistry. Telemed J E Health. 2018;24(8):639–648. doi: 10.1089/tmj.2017.0132 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref017] 17.Vinayagamoorthy K, Acharya S, Kumar M, Pentapati KC, Acharya S. Efficacy of a remote screening model for oral potentially malignant disorders using a free messaging application: A diagnostic test for accuracy study. Aust J Rural Health. 2019;27(2):170–176. doi: 10.1111/ajr.12496 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref018] 18.Haron N, Zain RB, Nabillah WM, et al. Mobile Phone Imaging in Low Resource Settings for Early Detection of Oral Cancer and Concordance with Clinical Oral Examination. Telemed J E Health. 2017;23(3):192–199. doi: 10.1089/tmj.2016.0128 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref019] 19.Lash T, Fox M, Fink A. Disease Misclassification, Corrections with Sensitivity and Specificity: Nondifferential and Independent Errors. In: Applying Quantitative Bias Analysis to Epidemiologic Data. Springer; 2009:94–96. [Google Scholar]

[pone.0270876.ref020] 20.Lyles RH, Tang L, Superak HM, et al. Validation data-based adjustments for outcome misclassification in logistic regression: an illustration. Epidemiology. 2011;22(4):589–597. doi: 10.1097/EDE.0b013e3182117c85 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref021] 21.Shaw PA, Gustafson P, Carroll RJ, et al. STRATOS guidance document on measurement error and misclassification of variables in observational epidemiology: Part 2-More complex methods of adjustment and advanced topics. Stat Med. 2020;39(16):2232–2263. doi: 10.1002/sim.8531 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref022] 22.Keogh RH, Shaw PA, Gustafson P, Carroll RJ, Deffner V, Dodd KW, et al. STRATOS guidance document on measurement error and misclassification of variables in observational epidemiology: Part 1-Basic theory and simple methods of adjustment. Stat Med. 2020;39(16):2197–231. doi: 10.1002/sim.8532 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref023] 23.Liao LJ, Chou HL, Lo WC, et al. Initial outcomes of an integrated outpatient-based screening program for oral cancers. Oral Surg Oral Med Oral Pathol Oral Radiol. 2015;119(1):101–106. doi: 10.1016/j.oooo.2014.09.020 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref024] 24.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. doi: 10.1016/j.oraloncology.2012.11.004 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref025] 25.Cheung LC, Ramadas K, Muwonge R, et al. Risk-Based Selection of Individuals for Oral Cancer Screening. J Clin Oncol. 2021;39(6):663–674. doi: 10.1200/JCO.20.02855 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref026] 26.Mallath MK, Taylor DG, Badwe RA, et al. The growing burden of cancer in India: epidemiology and social context. The Lancet Oncology. 2014;15(6):e205–e212. doi: 10.1016/S1470-2045(14)70115-9 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref027] 27.Kumar S, Debnath N, Ismail MB, Kumar A, Kumar A, Badiyani BK, et al. Prevalence and Risk Factors for Oral Potentially Malignant Disorders in Indian Population. Adv Prev Med. 2015;2015:20851 doi: 10.1155/2015/208519 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref028] 28.Sivakumar TT, Sam N, Joseph AP. Prevalence of oral potentially malignant disorders and oral malignant lesions: A population-based study in a municipal town of southern Kerala. J Oral Maxillofac Pathol. 2018;22(3):413–4. doi: 10.4103/jomfp.JOMFP_202_17 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref029] 29.Ramesh RM, Patrick S, Lotha Z, Azole Aier A, Birur NP. Prevalence and determinants of oral potentially malignant lesions using mobile health in a rural block, northeast India. Trop Doct. 2022;52(1):53–60. doi: 10.1177/00494755211049973 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref030] 30.Misra V, Singh PA, Lal N, Agarwal P, Singh M. Changing pattern of oral cavity lesions and personal habits over a decade: hospital based record analysis from allahabad. Indian J Community Med. 2009;34(4):321–5. doi: 10.4103/0970-0218.58391 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0270876.ref031] 31.Nadeau C, Kerr AR. Evaluation and Management of Oral Potentially Malignant Disorders. Dent Clin North Am. 2018;62(1):1–27. doi: 10.1016/j.cden.2017.08.001 [DOI] [PubMed] [Google Scholar]

[pone.0270876.ref032] 32.Warnakulasuriya S. Oral potentially malignant disorders: A comprehensive review on clinical aspects and management. Oral Oncol. 2020;102:104550. doi: 10.1016/j.oraloncology.2019.104550 [DOI] [PubMed] [Google Scholar]

PERMALINK

Smokeless tobacco use and oral potentially malignant disorders among people living with HIV (PLHIV) in Pune, India: Implications for oral cancer screening in PLHIV

Ivan Marbaniang

Samir Joshi

Shashikala Sangle

Samir Khaire

Rahul Thakur

Amol Chavan

Nikhil Gupte

Vandana Kulkarni

Prasad Deshpande

Smita Nimkar

Vidya Mave

Roles

Abstract

Introduction

Methods

Results

Conclusion

Introduction

Methods

Study setting and design

Study procedures

m-Health procedures

Oral Human Papilloma Virus (HPV) procedures

Images review procedures

Follow-up procedures

Fig 1. Scheme of study procedures.

Study definitions

Suspected OPMD (outcome variable)

SLT use (explanatory variable)

Statistical analysis

Results

Study population

Table 1. Demographic and clinical characteristics of participants recruited in an mHealth-based oral cancer screening study in Pune, India.

Prevalence of suspected and corrected OPMDs

Fig 2. Corrected prevalence of OPMDs under different plausible sensitivity and specificity parameters assuming non-differential misclassification of suspected OPMDs (i.e., SensitivityPLHIV = SensitivityHIV uninfected; SpecificityPLHIV = SpecificityHIV uninfected).

Associations of suspected and corrected OPMDs with HIV status and SLT use

Table 2. Prevalence ratios for suspected oral potentially malignant disorders among participants recruited in an mHealth-based oral cancer screening study in Pune, India.

Fig 3. Corrected adjusted prevalence ratios for OPMDs under different plausible sensitivity and specificity parameters assuming non-differential misclassification of suspected OPMDs.

Fig 4. Adjusted prevalence ratios for suspected OPMDs among current smokeless tobacco users by HIV status and duration of use.

Discussion

Conclusion

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Jitendra Kumar Meena

Roles

Author response to Decision Letter 0

Decision Letter 1

Jitendra Kumar Meena

Roles

Acceptance letter

Jitendra Kumar Meena

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Fig 2. Corrected prevalence of OPMDs under different plausible sensitivity and specificity parameters assuming non-differential misclassification of suspected OPMDs (i.e., Sensitivity_PLHIV = Sensitivity_{HIV uninfected}; Specificity_PLHIV = Specificity_{HIV uninfected}).