Cancer Awareness Among Lebanese Adults, A Cross Sectional Study

is the of in the of could be by reducing to risk factors. We explored population awareness of cancer risk factors, predictors of higher and to what awareness correlates with behaviour.


Abstract Background
Cancer burden in Lebanon is among the highest of countries in the Eastern Mediterranean Region, but around one third of cases could be avoided by reducing exposure to avoidable risk factors. We explored population awareness of cancer risk factors, predictors of higher awareness and to what extent awareness correlates with behaviour.

Methods
In 2019, we performed a cross-sectional study (N=726) that consisted of a questionnaire administered in face-to-face interviews.

Results
On spontaneous recall, pollution and smoking were the most mentioned causes of cancer. On average, 75% of risk factors were correctly recognized, 22% of non-carcinogenic factors, and 75% of protective factors. Woman and older people had better cancer risk awareness. Several non-carcinogenic factors are seen as more potent carcinogens than truly carcinogenic factors. There was a moderate correlation between awareness and behaviour for waterpipe, alcohol and physical activity. There was no such correlation for cigarette smoking.

Conclusion
Awareness campaigns are important, but need to be embedded into national behaviour change programs so that other barriers to behaviour change are also tackled.

Background
Lebanon has the highest cancer incidence rates of the countries in the Eastern Mediterranean Region and is in the top quartile of cancer incidence worldwide according to GLOBOCAN 2018. [1] The precise place in the ranking has been a source of debate, because compared to other countries in the region Lebanon was the rst to have early detection programs in place and has a better case registration system. Data from the Lebanese Cancer Registry 2016 show that the most common cancers among males are: prostate (17.0%), lung (14.4%), bladder (11.1%) and colorectal (10.1%). For females they are breast (39.1%), colorectal (9.0%), lung (6.4%), and non-Hodgkin lymphoma (4.6%). These ve locations account for more than half of all cancers (52.7% for males, 59.0% for females).
Further increase in cancer burden in the coming years is likely as the proportion of the elderly within the Lebanese population is increasing. [2] However, this trend can be slowed down: it has been estimated that around 33% of all cancers in adults in the Eastern Mediterranean Region region are attributable to eight avoidable lifestyle factors such as tobacco, alcohol, low physical activity, and high Body Mass Index (BMI). [3][4][5] Strategies to change existing behaviour are therefore essential to lower cancer burden.
Unfortunately, the prevalence of these lifestyle risk factors is increasing rather than decreasing. [2,4,6] To obtain a change in existing behaviour, increasing knowledge of risk factors and symptoms (summarized as "Cancer awareness") is necessary [7]. In Lebanon, data on cancer awareness is scarce but points towards a lack of awareness. [8] The rst objective of this paper is to analyse awareness of risk factors in the general Lebanese population and nd predictors of increased awareness. Secondly, we will assess whether the awareness of four selected risk factors (physical activity, smoking cigarettes, smoking waterpipe and alcohol consumption) is correlated to behaviour for that factor.

Methodology
We performed a cross-sectional survey in the general adult population in 2019. The survey consisted of a questionnaire administered during face-to-face interviews. A non-probability sample (n=726) was frequency matched to national government estimates on age, level of education and gender. Sampling took place in sampling stations of varying socio-economic characteristics in the Beirut and Mount-Lebanon governates. Inclusion criteria were Lebanese nationality and age ≥20 years old. Exclusion criteria were severe hearing or speaking di culties. All refusals were registered. No nancial incentives were provided.

Ethical considerations
Informed consent was sought from every participant by explaining the objectives of the study, that participation was voluntary, that anonymity of the respondents was guaranteed and that withdrawal at any point was possible. The study was approved by the ethical committee of the University Hospital of Brussels.

Questionnaire
The questionnaire (supplementary le 1) was based on two existing English language validated measures of cancer awareness (the Cancer Awareness Measure and the Health Information National Trends Survey), with some adaptations for culture speci c elements (such as the use of waterpipe). The questionnaire was translated to Modern Standard Arabic using forward-translation and back-translation, and then piloted on ve persons after which some clari cations were made.
The questionnaire collected data on chronic health conditions, lifestyle behaviour (including physical activity, alcohol and tobacco consumption), socio-demographic characteristics, preferred source of healthcare advice (open question), and risk factor awareness. Risk factor awareness was evaluated as "risk factor recall" and "risk factor recognition".
Recall was evaluated with an open-ended question: 'What do you think is the biggest risk factor for developing cancer?'. Recognition was measured by presenting 14 factors that increase cancer risk (see table 3), seven that have no in uence on cancer risk (see table 3) and three factors that lower cancer risk (see table  4), according to the International Agency for Research on Cancer (IARC). We then asked, 'Do you think any of these can in uence someone's risk of developing cancer?' There were eight response options: 'increases' (three options: strongly, moderately, somewhat), 'neither increases nor decreases', 'decreases' (same three options) and 'I don't know'.

Sample size and power calculation
Epi-info was used to calculate the required sample size, using the following equation: where Z is a standard normal variate (Z1-α/2= 1.96 at 95% con dence interval), d is the absolute accuracy or precision (5% marginal error), P is the expected proportion of the population with a speci c outcome and was set at 0.5 (the advised value if the proportion in the population is not known). This yielded a necessary sample size of 385, which was doubled to compensate for a design effect.

Awareness of risk factor
For the factors that increased risk and the factors that have no effect on cancer risk, we calculated two indicators: Correct Recognition Percentages & Carcinogenic Strength Score. The rst was the proportion of people that recognised that factor correctly (all three "increase" options were seen as correct for the risk factors, and the option "neither increases nor decreases" was seen as correct for the factors that do not in uence risk). The Carcinogenic Strength Score quanti ed how strongly people thought that particular factor increased the cancer risk. The option 'Strongly increases' was given value 10, 'Moderately increases' got value 6.66, 'Increases somewhat' got value 3.33, all other options got value 0. The Carcinogenic Strength Score of a given factor was the arithmetic mean of the individual scores.
For the factors that decrease risk, we only calculated the Correct Recognition Percentages (by seeing all three "decrease" options as correct).

Predictors of awareness
To nd predictors of awareness, we grouped the three types of factors (increasing, decreasing, no effect) so that each individual had a three Grouped Correct Recognition Percentages. We then ran multiple linear regressions to determine independent predictors of these three percentages. Variables were candidates for purposeful selection in a multivariate regression if their univariate correlation had p-values <0.2. We tested demographic variables (age, level of education, monthly income), chronic illness variables (diabetes, hypertension, arthritis, chronic lung disease, heart condition) and variables on cancer experience (affected self, partner, family). Whether or not a variable was withheld in the model was based on the likelihood ratio test with signi cance set at p<0.05 (age and gender were always included as adjusting variables).

Correlation between awareness and behaviour
For waterpipe, cigarette, alcohol and physical activity, we assessed the association between perception and behaviour. The behaviour answers for those four were categorised and used as dependant variables in an ordered logistic regression that had the Carcinogenic Strength Score as independent variable (signi cance set at p<0.05). This regression was always adjusted for gender and age since these two variables have previously been shown to be associated with lifestyle behaviour. We tested the proportional odds assumption with the user-written command called omodel, which does a likelihood ratio test, with signi cance set at p<0.05.

Results
Of the 861 people that were approached, 123 refused to participate (14.3%) and 738 completed the survey (85.7%). Of these 738, inclusion and exclusion criteria were met by 726 (98.5%). Table 1 shows socio-demographic characteristics of the sample besides national data. The distribution of gender, age, civil status, and education is similar between sample and general population. The sample has a higher proportion of employed people (62.3% versus 48.7%).

Awareness of risk factors on recall
In the recall question, pollution (40.0%, N=289) and smoking (35.0%, N=253) were perceived as the most important causes of cancer (see table 2). There were no signi cant differences between age, gender, or education status.

Awareness of risk factors on recognition
The Grouped Correct Recognition Percentages were: 75% for carcinogenic factors, 22% for the factors without in uence on cancer risk, and 75% for the protective factors. The highest Correct Recognition Percentages were attained for cigarette smoking, waterpipe smoking and air pollution (97.5%, 96.6%, and 94.6% respectively, see table 3).
The highest Carcinogenic Strength Score (table 3) was measured for smoking waterpipe (8.2), closely followed by smoking cigarettes (8.1) and air pollution (8.0). The factors without in uence (marked between "" in table 3), had Carcinogenic Strength Scores as high as 6.1 (Food colouring), which was higher than some of the truly carcinogenic factors: sedentary lifestyle (2.4) and HPV infection (2.3).
The correlation between awareness and behaviour Table 4 shows the results of the regression of awareness on behaviour. Each step of seeing waterpipe as more harmful was associated with a lower frequency of waterpipe use (OR up 5.51, p=0.001). The same was found for alcohol (OR up to 3.66, p=0.020) and the bene ts of physical activity (OR up to 3.26, p=0.008). There was no such correlation for cigarette smoking.

Discussion
Eight avoidable lifestyle factors are responsible for a third of all cancers in Lebanon. In order for people to change their behaviour (stop smoking, drink less alcohol, become more physically active, etc) they need to be aware of the risks of these lifestyle factors.
The most often correctly classi ed factors were smoking cigarettes (97.5%), waterpipe (96.6%) and air pollution (94.6%), while the least known factors were HPV (32.6%), sedentary lifestyle (48.6%), alcohol (63.4%), and being overweight (65.8%). This order in awareness of factors is similar to the one obtained in a WHO survey. [12] Recognition is highest for protective factors and for factors that increase cancer risk, while factors that do not in uence cancer risk were often falsely seen as carcinogenic by a large percentage of people (food colouring: 93.7%; tap water: 83.9%; depression and anxiety: 76.6%). Food colouring, depression and tap water also had high carcinogenic perception score (6.1, 4.5 and 3.9 respectively), which means they were perceived as more carcinogenic than half of the surveyed factors that truly are carcinogenic. These results may re ect the success of past Lebanese awareness raising campaigns, but also the fact that such campaigns often fail to su ciently address cancer myths. Attributing a carcinogenic effect to neutral factors (believing in cancer myths) has also been observed in other countries. In the UK, exposure to stress is seen as carcinogenic by 43% of adults, food additives by 42%, and mobiles phones by 26%. [27] It seems likely that new media such as social media are partly to blame for this. [27] We found young adults and men to have lower awareness, which con rms results from previous studies. [9,10] Somewhat unexpectedly, neither income nor education were correlated with awareness, although these have been shown to be correlated in other countries. [10] A possible explanation could be that cancer is still considered a taboo in Lebanon, in every layer of society. [11] There was a moderately strong correlation between awareness and behaviour for waterpipe, alcohol and physical activity, indicating that raising awareness might change behaviour for these factors. There was no such correlation for cigarette smoking, meaning other strategies than awareness building must be used to change behaviour.

Information material of IARC (www.euro.who.int, European Code against Cancer) is already available in
Modern Standard Arabic and could be used by local projects with minimum need for funds. These materials could be altered centrally to include information on cancer myths. However, raising awareness is not enough. A national behaviour change program should be designed, with multifaceted interventions that are sustained for many years. We also found this in our results: although better understanding of the dangers of waterpipe, physical activity and alcohol were correlated with more bene cial individual behaviour for those factors, this was not the case for cigarette smoking. Previous research also found a large gap between knowledge, intention and behaviour, and concluded awareness is not su cient to change behaviour. [14,29,30] In 2011, Michie proposed a framework named the Behaviour Change Wheel, which includes eight intervention categories and 7 policy categories that need to accompany awareness campaigns. [7] Smoking Prevalence of this risk factor among Lebanese adults is 38.0% (cigarettes) and 36.9% (waterpipe). [13] Cigarette smoking is more common for men (47.6%) than for women (29.0%). Cigarettes are responsible for 24.4% of cancers in men and 5.2% in women. [3] Cigarettes and waterpipe were identi ed as harmful by over 95% of respondents and were given the highest carcinogenic perception scores. [8] Although water pipe smoking has similar health effects as cigarette smoking, its harmful effect is still neglected in anti-smoking campaigns. [15] In neighbouring Jordan, a similarly high awareness was found (68.4% of women see smoking as the most important risk factor for ovarian cancer, 78.4% of Syrian refugees see smoking as a major risk for cancer). [16,17] Air pollution Lebanon has an average PM2.5 concentration of 24.3 μg/m³ (PM2.5 should not exceed 10 μg/m³ annual mean), while in urban areas air pollution can go up to 30.3 μg/m³. More than 93% of this is due to road tra c [18]. PM2.5 air pollution is responsible for 1.7% of cancers in men and 0.7% in women. [3] It is therefore reassuring that air pollution was identi ed as harmful by over 95% of respondents, but slightly troubling that it is seen as the number one cause of cancer in the recall question, and was given the same Carcinogenic perception score as Tobacco consumption. Tobacco accounts for 15.1% of all cancers in Lebanon, and its relative risk is considerably higher than that of 24.3 μg/m³ PM2.5 air pollution [3]. This means the individual effect of smoking cessation is much bigger than that of avoiding air pollution.

HPV
Prevalence of high-risk HPV type 16-18 in Lebanese women is estimated to be around 3% (95% CI 2.1%-4.3%). Among Lebanese women in 2016, 1.9% of cancers are cervical cancers, which are nearly all caused by HPV. This means at least 2% of cancers among Lebanese women are attributable to HPV, not considering other locations of HPV related cancers. [19] It is therefore a concern that HPV is hardly known as a cause of cancer: 32.6% of respondents correctly identi ed HPV as a cause of cancer, with a Carcinogenic Perception score of 2.3, the lowest score, even lower than all the harmless factors. This con rms ndings of previous studies, that found that 34% of adults were aware of the cancer risks of HPV, and 18.2% of adults were su ciently educated concerning HPV infection. [20][21][22] Sedentary lifestyle Prevalence sedentary lifestyle among Lebanese adults is high: 60.7% of men and 61.3% of women do not meet WHO recommendations on physical activity. [13] Physical inactivity is responsible for 0.6% of cancers in men and 3.2% in women. [3] This percentage is likely to be higher since the prevalence of physical inactivity used in that study (36.0%), has since been adjusted upwards (61.0%). [23] Only 48.6% of people correctly identi ed physical inactivity as harmful. Previous Lebanese research (n=371, self-administered questionnaire) found similar results: 36.5% identi ed a sedentary lifestyle as a risk factor for colorectal cancer. [24] Knowledge about this risk factor is better than in certain other settings (only 16.6% of refugees in Jordan identify sedentary lifestyle as a risk factor), but still insu cient [16].
Alcohol Regular alcohol consumption (at least once in the last month) has a prevalence of 23.4% among Lebanese adults (28.1% among men and 18.9% among women). [13] Alcohol is responsible for 1.1% of cancers in men and 1.5% in women. [3] This percentage is likely to be higher since the prevalence of alcohol consumption used in that study has since been adjusted upwards (60.2% of adults used alcohol at some point in life versus a previous estimate of 9.9%). [23] Alcohol was known to be carcinogenic by 63.4% of people, with a Perceived Carcinogenic Score of 3.5. This makes alcohol one of the lesser known carcinogenic factors in Lebanon. Previous research had similar results: in a self-administered questionnaire in Lebanon 52.8% identi ed alcohol as risk factor for colorectal cancer, and in a population-based survey in Iran 56% identi ed alcohol as a risk factor. [24,25] High-BMI Being overweight (>=25kg/m 2 ) has a prevalence of 64.9% among Lebanese adults (71.3% among men and 58.9% among women). [13] High BMI is responsible for 2.7% of cancers in males and 8.1% in females. [3] This puts high BMI as the second most important avoidable cause of cancer for women in Lebanon.
High BMI was identi ed as a risk for cancer by 65.8% of respondents, with a Carcinogenic perception score of 3.7. In comparison to other factors, that is much lower. A previous study in Lebanon showed that 51.4% consider high BMI a risk factor for cancer. [26] Obesity is also poorly recognised as a risk factor in England. [27] Strengths And Limitations Our population-based sample was frequency matched to important demographic variables. This method is more likely to give representative results compared to a non-random sampling method that does neither check nor adjust for population distribution. We used a modi ed validated measure of cancer awareness, managed to also reach the lower socio-economical groups and had very few missing data except for the variable personal income.
Our study also has limitations. Recall and recognition are different methods of measuring cancer awareness, recall is thought to underestimate awareness, while recognition is thought to overestimate it. [10,28] Of the 8 Lebanese governorates, our sample strategy included only two (mainly urban). Therefore, it cannot be guaranteed that our results are generalisable to Lebanon as a whole, although our sample was matched to national data on socio-economic variables.

Conclusion
The potential for cancer prevention is high in Lebanon: one third of all cancers could be avoided by changing behaviour. Several conditions need to be met to achieve the necessary behaviour changes. The population needs to be aware of the risks of their own behaviour and also be informed about cancer myths. However, such awareness campaigns need to be embedded into national behaviour change programs so that other barriers to behaviour change can also be tackled, this seems especially true for cigarette smoking. The study was approved by the ethical committee of the University Hospital of Brussels with reference number CAL-3.1. All procedures performed involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was sought from every participant by explaining the objectives of the study, that participation was voluntary, that anonymity of the respondents was guaranteed and that withdrawal at any point was possible.

Consent for publication
Consent for publication was not be required since we do report on individuals in the manuscript.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.