Dental caries experience in adults: a cross-sectional community survey within Ethiopia

Background Ethiopia is a large and diverse developing sub-Saharan African country with increasing prevalence of non-communicable diseases, including oral conditions. Oral health and dental care have been given little consideration, and there is limited information relating to population oral health and use of dental services in the country. The aim of this study was to examine the burden and impact of dental caries experience and investigate access to dental care amongst adults within Ethiopia. Methodology This community-based oral health survey is a baseline study for the ASSET - Health System Strengthening in sub-Saharan Africa project undertaken in the Butajira area, south-central Ethiopia. A stratied random sample of households and individuals were participated in the study. The survey was informed by the World Health Organization (WHO) Oral Health Survey Methods (5th ed.) at International Caries Detection and Assessment System (ICDAS) threshold 4 and the UK Adult Dental Health Survey 2009. Face-to-face interviews and clinical dental examinations were conducted. The data were analysed using descriptive statistics; and a binary logistic regression model was built to assess the association of dental caries and predictor variables: demographics, health behaviours, quality of life, dentition status and consequences of untreated caries data in adults ( ≥ 18 years). Just and many never attending a dental service. Sixty percent the adults had dental caries, whom untreated carious teeth. fully adjusted logistic regression age, dental Khat positive signicant with dental caries whilst education random 2

to participate in the study and provide written consent. People with serious health conditions, making it di cult to comply with the interview and oral examination, were excluded from the study.

Instruments and Measures
The 5th edition of World Health Organisation (WHO) Oral Health Survey Methods (5th ed.) (28) formed the basis for the survey methodology and the standard instruments (electronic questionnaire and data collection forms) were developed drawing on the latest epidemiological research methods: the UK (England, Wales and Northern Ireland) adult dental health survey 2009 (29), and the International Caries Detection and Assessment System (ICDAS) dental caries scoring system (30). Dental caries experience at tooth level, and the consequences of untreated caries are the main dependent variables. Dental caries experience was determined using the globally accepted DMFT index (the presence of Decayed, Missed and Filled Teeth) extracted from the dentition status data (28). Assessment of dental caries was based on the WHO survey guideline 'tooth condition' assessment format (28), and the threshold of individual lesions was determined using the International Caries Classi cation and Management System (ICCMS), ICDAS coding criteria. We considered ICDAS 4 (D 4 ), and above as 'caries', because ICDAS 3 (D 3 ) and below is more di cult to assess when only visual examination is used without clinical/diagnostic facilities (30), particularly in areas where Khat chewing is common, as it can stain teeth after repeated use making early caries mode di cult to identify. Khat, a green leaf chewed for its mild stimulating effect, is commonly used in East-Africa and Middle-east region (31,32).
To determine the consequences of untreated caries, we used the PUFA index, presence of Pulpal involvement, Ulcer, Fistula and Abscess (PUFA) (33). We also assessed the presence of pain or any condition related to the chronic complication of dental caries as an additional outcome variable (29).
Sociodemographic factors included in the baseline surgical questionnaire instrument as important independent variables for subsequent analysis were as follows: age, sex, education status, religious a liation, marital status and location (urban/rural) as these variables are common predictors of caries (34,35). In relation to lifestyle and behavioural factors, frequency of tooth cleaning and materials used, including toothpaste were collected in the dental data collection instrument and used as proxies for oral hygiene/plaque removal and uoride exposure (36,37). Other health behavioural and lifestyle factors examined because of their in uence on oral health were: Khat chewing, sugar, tobacco and alcohol consumption (9,38). Self-reported dental attendance, which refers to attending a recognised dental service provider, was also obtained to look at access to dental care and possible barriers to care (14,39).

Standardization
The dental questionnaire instrument was translated into the working language of the area -Amharic. It was also back translated into English by two native Amharic speakers with full professional competencies in English. They were non-dental/medical professionals with no knowledge about the survey. The original and translated questionnaires were then assessed if they achieve semantic and conceptual equivalence, and it did not require any further amendment.
The clinical examiners (dentists) were trained over a two-day period and calibrated for identi cation and assessment of dental caries on the same set of teeth. Inter-examiner agreement was computed using the kappa statistics, with a range of 0.6-1.0 (mean: 0.96). A standard form was created online for recording of the clinical data. Repeat examinations were not possible because of time and logistical constraints.

Data Collection
Lay data transcribers and dentists teamed up with their surgical community survey counterparts during the data collection. Participants were interviewed face-to-face and examined for oral diseases in their respective households. The transcribers recorded all information (questionnaire and clinical oral examination) using the Open Data Kit (ODK) software on smart phones. Oral examination was visual, conducted using standard disposable dental mirrors and blue-white spectrum solar headtorches with the examinee supine. Dental probes were only used for removing debris, when required.

Data Entry, Cleaning and Analysis
The ASSET data specialist exported the data and merged the surgical and dental data using unique identi er codes to form the datasets for analysis. We then cleaned the dental data removing any incomplete cases related to this analysis, created new variables and undertook data analysis using SPSS version 25. The numerical variable-age was recoded into six age-bands (28), and the variable dental caries experience was categorized in to D 4 MFT = 0 and D 4 MFT ≥ 1 prior to the data analysis. Sugar consumption 'risk' was also calculated based on the frequency of reported intake of sweets, cakes, biscuits and soft/ zzy drinks. Consuming at least one of them more than once a week was categorised as 'higher-risk', between once a month and once a week as 'moderate-risk', and rarely or never as 'lowerrisk' consumption.
The sociodemographic characteristics, health behaviours and other relevant data were summarized using descriptive statistics. We used frequencies and proportions to describe the variables and used a chisquare test to examine the association between the independent and dependent variables outlined above. The association of health behaviours, tooth cleaning habit and the sociodemographic factors with dental caries experience were further analysed using both unadjusted and adjusted binary logistic regression models.

Results
Of the 1071 people who participated in the study (89.3% response rate), 650 were adults. Twenty-four adult participants (3.8%) were excluded from the nal analysis because of missing relevant information. Therefore 626 participants, with an average age of 38.4 years (range: 18-100; sd: 16.4 years) were included for analysis. A higher proportion of the study population was female (63.9%, n = 400), married (71.4%, n = 447), and Muslim (76%, n = 476). Just over half (53.2%, n = 333) lived in rural areas and 44.4% (n = 278) had not received any formal education.
Reported health behaviours were largely positive. The vast majority (97.9%, n = 613) reported never smoking cigarettes or drinking alcohol (90.4%, n = 566), or chewing Khat (59.0%, n = 370); sugar consumption was generally low with only 10.7% (n = 67) in the 'higher-risk' category (consuming any sweet more than once a week). However, tooth cleaning twice a day or more was only practiced by about one in ve adults, as was toothpaste use. Chew-sticks (Mefakiya) were most commonly used for tooth cleaning (45.0%, n = 281).
A majority (74.0%, n = 463) of participants had never been to a dental service. The main reported reason for not attending a dental service was never experiencing any dental pain/problem (71.3%, n = 330), followed by the distance to a dental clinic/service (9.3%, n = 58) and the cost of treatment (8.2%, n = 51). Only 37.1% (n = 140) of those with dental caries experience reported attending a dental service.
Dental caries experience D 4 MFT and its consequences in relation to demographic and behavioural variables are presented in Table 1. Six out of ten (60.2%, n = 337) of adults surveyed had a D 4 MFT score of 1 or more, out of which 88.0% (n = 332) had at least one untreated obvious carious tooth (D 4 T) which accounts for 53.0% of the total study population.    4 MFT scores ranged from 0-28 (mean, 4.4 ± 6.9), and the score increased with age (Fig. 1). Only 0.8% (n = 3) of adults with dental caries experience had llings, but 49.9% (n = 188) of them had missing teeth.
Pain and/or discomfort was reported by 16.5% (n = 103), with 7.2% (n = 45) having one or more PUFA component. Most (59.9%, n = 226) adults with dental caries experience reported some tooth pain and/or discomfort during the last year. Based on the bivariate analysis, increasing age had a signi cant and positive association with dental caries experience; whereas educational level was negatively associated. Marital status and location also demonstrated highly signi cant associations with caries (p < 0.001 for all). However, no association was observed with sex and religion. Regarding the health behaviours, dental attendance, frequency of tooth cleaning and materials used (including the use of toothpaste), sugar intake and Khat chewing were highly associated with dental caries experience (P < 0.001 for all). There were also signi cant associations between the health impact, notably tooth pain/discomfort related to untreated dental caries and age, education, marital status, dental attendance, location and tooth cleaning frequency (p < 0.05). Out of all the independent variables, only dental attendance, tooth cleaning frequency and materials used for dental hygiene were signi cantly associated with PUFA (p < 0.05).
The binary logistic regression models for dental caries experience and predictors are presented in Table 2.
In the unadjusted model, all the independent variables, except for sex, had signi cant associations with dental caries experience (p < 0.05). In the adjusted model, the odds of experiencing dental caries were higher in older age-groups, whilst the odds of dental caries experience were lower in higher levels of education. The odds of experiencing dental caries amongst those who reported attending dental services were seven times (95%CI: 3.9-12.1, p = 0.001) the odds of those who had never attended. Khat was the only lifestyle behavioural component that remained signi cant in the adjusted model; people who reported chewing Khat daily or nearly every day had four times (95%CI: 1.7-8.3, p = 0.001) the odds of experiencing dental caries as compared with reported non-chewers. Location of residence and other health behaviours including tooth cleaning, the risk of sugar intake, did not predict caries experience after adjusting for other factors.

Discussion
In this cross-sectional study amongst a rural and urban population of south-central Ethiopia, six out of ten adults were found to have dental caries experience. This study is one of the few oral health research projects to have been carried out in the country and the rst population-based oral health survey to be part of a large multi-disciplinary global health project providing an overview of adult oral health in the community and the most extensive in nature. This research has certain limitations which need to be acknowledged. First, it was conducted in a relatively ethnically homogenous population, which therefore cannot be considered nationally representative. Second, almost half (46.8%) of the study participants were from the urban area due to the accessibility challenge in some rural areas. Since a large proportion (80.0%) of the general population in the country is rural, this again needs to be recognised as unrepresentative of the national situation; however, it allowed comparison between the rural and urban settings and given the study ndings whereby the differences between urban and rural settings were not signi cantly different in the adjusted logistic regression model, this may not be a major factor. Third, the people who stayed at home and were available to participate in the data collection might be different from those who were out engaging in work and other activities and thus possibly overrepresented. However, every effort was made to ensure that people who worked away from home were included in the study, although most men working in the elds might have been missed. Fourth, the questionnaire data were collected by face-to-face interview, which means participants may have given more socially desirable answers about their lifestyle and behaviours. None-the-less, given the very low level of education in this community, this was the most appropriate survey method. Fifth, the sample size was calculated based on surgical care needs; however as surgical care needs are less common than oral diseases, this provided a substantial sample for the oral health survey when compared with other national dental surveys (29), and our ndings will provide a sound basis to inform future epidemiological studies in Ethiopia.
The magnitude of untreated obvious caries (clinically visible lesions representing demineralisation in the middle third of the dentine, and above indicated by the ICDAS 4 threshold) in the study participants was high; if the early and sub-clinical lesions were included, which would require optimal clinical and radiographic examinations (40), a higher value would result. These ndings suggest an increasing burden of NCDs in general (41). Interestingly, a higher proportion of dental caries experience was observed in rural areas, unlike a previous study in Ethiopia which reported a higher prevalence of caries in urban areas (61.1%) (14). However, the odds of experiencing dental caries were almost the same for both locations in the adjusted logistic regression model. Although rural areas are generally suggested to have less access to cariogenic foods, this may be changing with the nutrition transition in Africa; in addition to the limited access to dental care and preventive services as compared to urban areas. Similar issues exist in both lower and higher income countries, whereby the rural population has more untreated caries (35,42).
In the study participants, the odds of experiencing caries increased with age and the pattern of the condition is also cumulative throughout life by its nature (34). This nding mirrors the emerging result of the Dunedin longitudinal study (the 1972/3 birth cohort) of dental caries trajectory assessment in which the participants dental caries experience in the different categories of the trajectory increased with age (43). However, there was no signi cant difference between males and females in our study regarding their dental caries experience unlike previous studies in the UK (44), but it is similar with the recent global burden of oral health study ndings (6,45).
In this regression model, the odds of experiencing dental caries signi cantly reduced with increasing educational attainment of adults. Education, being one of the social determinants of health can also in uence an individual's position in the society and determine their income. This in turn in uences the risk of one's exposure to different diseases (25), including dental caries in adults (34).
Health behaviours, whilst important risk factors for oral disease (9), are less likely to be signi cant in our model. Amongst the health behaviours assessed in this survey, chewing Khat daily was associated with increased odds of dental caries experience when keeping other variables in the model constant. The association of the oral hygiene practice and sugar consumption was no longer present after the statistical adjustment. However, people who reported cleaning their teeth twice or more per day were those with a higher dental caries experience. This might be explained by the fact that those who already have problem tend to perceive severity of the disease and clean their teeth more (46), but do not appear to be using uoride toothpaste (47). Khat chewing has been suggested to be associated with some oral conditions including periodontal diseases, attrition of the tooth surface and tooth discolouration, oral cancer and other mucosal diseases. According to Al-Maweri (2018), Khat itself is not cariogenic. However, it has a dehydrating effect on the oral mucosa and a bitter taste. Consequently, Khat chewers tend to consume a large quantity of non-alcoholic uids, such as carbonated soft drinks, water, and coffee (usually sweetened). Some also supplement the leaves with re ned sugar and it also increases the tendency to tobacco smoking (38). A cohort study of 98 Yemeni Khat chewers and 101 non-chewers, aged 18-35 years with early occlusal caries suggested a higher risk of dental caries progression in lowincome Khat chewers as compared to the non-chewers, although the study follow-up was short (48).
Whilst dental attendance was generally uncommon, it was notable in those with dental caries experience in our study, only just over one-third had attended a dental service. However, those who had been to the service had seven times the odds of dental caries experience than those who never did. This might suggest that dental visit was perceived need driven and given the level of untreated diseases, care was probably limited to the offending lesion. In this regard, the main reasons for not attending dental services are generally multifactorial (39), and it can also be highlighted by the barriers of access related to the direct and indirect costs. A study of Jordanian adults on factors in uencing access to dental care also reported the cost of treatment and perceived lack of treatment need, following lack of time as main reasons for not attending a dental care (49). However, our study participants did not highlight lack of time as a reason, but mainly distance to the dental service and cost of treatments. Another study among adult city-dwellers in Burkina Faso also highlighted cost of dental treatment as the main barrier to dental visits (50). Dental treatment is known to expose households to unpredictable and signi cant nancial costs; recent research has shown, in low income households the larger portion of their disposable income could be used for dental treatment mainly in out-of-pocket payment situations (51). This could also be the case in Ethiopia, as out-of-pocket payment is the main method of funding healthcare (52).
According to our ndings, half of the people with dental caries experience had missing teeth due to caries, and less than one percent had dental llings. This indicates that tooth extraction is the most widely used treatment modality for dental caries in the area. This can be explained by late presentation with disease, due to challenges in access to care resulting in limited choice of treatment. However, socioeconomic conditions can also determine the access to preventive and restorative dental treatments (53), in addition to the inequity of healthcare services which is a challenge in most developing countries. However, the use of primary care services helped most industrialized countries to achieve greater equity so that people with greater health needs can get better access to services (54), albeit with less regularity. Reviews of oral health in the African region also suggested those challenges can be addressed by creating universal and equitable access to quality, affordable and appropriate oral health services (55).
This research provides important insights to dental caries experience in Ethiopia and provides pilot data to inform a nationally representative survey of oral health. This will help to inform appropriate workforce capacity building, through modelling future scenarios for dental service provision. It will be important to develop research informed feasible ways of addressing their oral healthcare needs of urban and rural populations to inform health policy on health promotion and healthcare services. Further research should also investigate oral health in relation to NCDs, together with the possible relationship of Khat and dental caries.

Conclusion
The nding suggests that there is a high burden of dental caries amongst adults in the district of Butajira and marked social inequity. Caries experience increases with age and is less prevalent amongst educated and urban populations. Generally, dental attendance is low and there is a high need for dental care.
Overall, this implies a clear need of health system strengthening interventions including oral health promotion.

Declarations
Funding and Disclaimer