Predictors of herbal medicine use among cancer patients

Moroccan studies have shown a high frequency of use of traditional and complementary medicines (T&CM). No survey has investigated the predictive factors of the use of traditional herbal medicine. This study aims to assess the prevalence, the determinants, and the predictors of this use. At the regional oncology center of Beni Mellal, we conducted a descriptive and analytical cross-sectional study with 530 adult cancer patients chosen by quota sampling. A 54-item questionnaire was used to perform the face-to-face survey. The Complementary and Alternative Medicine (CAM) healthcare model was used to create this instrument. The prevalence of traditional herbal medicine use was 36.6%. Several variables of the conceptual model were associated with the use of herbal medicine (gender, attitude, and belief towards T&CM, financial resources, and clinical characteristics of patients) (p < 0.05). Predictors of traditional herbal medicine use were female gender (OR = 4.687; p < 0.000), social impact of cancer (OR = 2.314; p < 0.05), emotional impact of cancer (OR = 2.314; p < 0. 05), expected benefits of T&CM use (OR = 3.453; p < 0.000), perceived scientific barriers (OR = 0.580; p < 0.000), perceived socioeconomic barriers (OR = 0.657; p < 0.005), and social norms of relatives (OR = 3.238; p < 0.000). The use of herbal medicine is quite frequent. The profile of users is represented by women who experience pain and the psychosocial impact of cancer and who have positive beliefs about TCMs, are influenced by their relatives and who did not perceive the barriers of this use.


Introduction
Cancer patients in developing countries such as Morocco frequently use complementary and alternative medicine (CAM). Traditional medicine (TM) is synonymous with these complementary medications in Morocco (Brahmi et al. 2011). Traditional and complementary medicine (T&CM) combines the two terms TM and CAM and refers to the use of Health practices or products that do not fall under the purview of mainstream medicine and which have not been fully incorporated into the country's dominant health system (OMS 2013).
CAM incorporates natural products such as herbs, vitamins, mineral elements, and probiotics, while the group of mind/body practices includes procedures or techniques such as yoga, chiropractic, osteopathic manipulation, and acupuncture (Center for complementary medicine 2018).
T&CM use in cancer patients varies from 44.7 to 72% in Europe (Lüthi et al. 2021;Ciarlo et al. 2021;Bonucci et al. 2022). A recent systematic review analyzing data from 25 studies in developing countries involving 6878 cancer patients found that 54.5% of patients reported using T&CM for cancer care. Of these, 26.7% of users combined T&CM with conventional medicine. The prevalence in the Middle East and North Africa region was found to be slightly higher than in other regions (68%) (Hill et al. 2019).
A recent systematic review, which focused only on herbal medicine, including 155 studies and data from 809,065 participants, found that the prevalence of herbal medicine use by cancer patients worldwide was 22% (95% 1 3 confidence interval (CI) 18-25%), and the prevalence was higher in Africa than in other regions of the world (40%, 95% CI 23-58%) and in low-and middle-income countries (32%, 95% CI 23-42%) (Asiimwe et al. 2020).
The frequency of T&CM use continues to increase. Four recent studies in Asia, Europe, and Africa have confirmed this. For example, the prevalence was 61.2% in Malaysia and 82% in Taiwan (Chin et al. 2020;Hamed Abdalla et al. 2020). In Europe, almost 70% of Swiss pediatric oncology patients used a T&CM modality after diagnosis (Lüthi et al. 2021), and more than half of patients in Germany used at least one category of T&CM, the majority of which used dietary supplements (Dufter et al. 2021). In Africa, almost half of cancer patients are users of T&CM, including traditional herbal medicine. For example, 55.4% of cancer patients in Uganda reported using a T&CM modality (Mwaka et al. 2019). In North Africa, Tunisians with cancer used T&CM with a prevalence ranging from 40.54 to 85% in 2020 (Toukabri et al. 2020;Labidi et al. 2020).
In Morocco, cancer patients widely used medicinal plants as T&CMs, with a frequency ranging from 11 to 39% between 2011 and 2019 (Brahmi et al. 2011;Hessissen et al. 2011;Kabbaj et al. 2012;Tazi et al. 2013;Chebat et al. 2014;Samouh et al. 2019). Most of these studies conducted in Morocco were descriptive, and none of them specifically examined the specific predictors of the use of traditional herbal medicine in cancer patients. However, it is recommended to study the predictive factors of the use of each T&CM modality (natural products like herbal medicine and mind-body practice like yoga) (Patterson et al. 2002;Fouladbakhsh et al. 2005;Akyol and Öz 2011).
On the other hand, the existing literature review on T&CM use lacks theoretical modeling to explain this behavior and to compare results between different studies, as well as the identification of predictive factors regarding the purpose of use (differentiating between users for any the purpose of treating cancer or only improving quality of life) (Fouladbakhsh and Stommel 2007).
In addition, few studies have examined the psychological factors in cancer patients' use of herbal medicine. Indeed, international studies have focused on other socioeconomic and clinical factors (Engdal et al. 2008;Afifi et al. 2010;Guo et al. 2012;Ali-shtayeh et al. 2016;Alsharif and Mazanec 2019;Bazrafshani et al. 2019;The user et al. 2021). In Morocco, only one study looked at predictors of T&CM use in general, and the variables included in this study were sociodemographic variables, cancer stage, cancer treatment, and family history of breast cancer (Rahou et al. 2017).
As a result, the goal of this study is to assess the prevalence and determinants of traditional herbal medicine usage among cancer patients in the Beni Mellal-Khenifra region of Morocco, as well as the predictors of this use and the objectives of this use (sociodemographic, biomedical and psychological).

Population and study design
This is a cross-sectional analytical study conducted on a sample of cancer patients in the Beni Mellal-Khenifra region of Morocco. The approval of the study was obtained from the Ethics Committee for Biomedical Research of Oujda (CERBO) affiliated with the University Mohamed I under the number 29/2020. We recruited 530 cancer patients at the regional oncology center of Beni Mellal between February and July 2021. The sample size was calculated according to the following formula: where n is the sample size: the minimum size was 329, N is the population size = 2240 (2240 patients followed in the center at the end of 2020), z is the z score = 1.96 (the z score is the number of standard deviations of a given proportion from the mean, it corresponds to a confidence level of 95%), e is the standard error = 5%, and p is the standard deviation = 0.5. We used quota sampling by gender (70% women and 30% men) and cancer type to recruit participants. Cancer patients were recruited from different departments of the oncology center (radiotherapy, chemotherapy, day hospitalization, and medical consultation).
The participants came from all over the Beni Mellal-Khenifra region. They were at least 18 years old and had a confirmed diagnosis of cancer. These participants were selected from patients who were able to answer the questions and who agreed to participate voluntarily in the study. We excluded patients who were too tired to respond. We contacted patients in the waiting room of the chemotherapy department, the waiting room of the radiotherapy department, and in the hospital rooms. Each patient was interviewed face-to-face after their agreement.

Conceptual model
The model used in this research is the CAM healthcare model (Fouladbakhsh and Stommel 2007), which extended Anderson's socio-behavioral model. This model examines the concurrent use of conventional medicine and T&CM.
In addition, this model allows researchers to study T&CM use separately according to the purpose of use, for example, the use for definitive treatment of the disease or to improve quality of life and promote health (Fouladbakhsh and Stommel 2007;Lorenc et al. 2009). The CAM Healthcare model aims to improve the understanding of the health phenomenon of T&CM use to predict use, objectives of use, and outcomes of use. This model includes all the factors identified in the Anderson Sociobehavioral Model and enhances that framework by making it specific to T&CM use by adding independent variables to the T&CM. A dependent variable allows for the separation of T&CM practices and products.
This allows us to explain the use of T&CM, in this case, herbal medicine, according to the following variables: Predisposing variables The patient is predisposed to use T&CM; these are sociodemographic and psychological variables such as ethnicity, cultural and community practices, beliefs, and attitudes towards traditional medicine.
Enabling variables Factors that facilitate or limit the use of health services include resources specific to individuals and families that could influence this use, such as income, health insurance, occupation, region of residence as well as an urban or rural area of residence, availability of health services such as accessibility and proximity of health services, physician to resident ratio, and litter capacity.
Need variables The individual's need for care, the experience of illness, and perceptions of health.
The dependent variable in this model is healthcare utilization. This concept from Anderson's model was taken and modified by the CAM healthcare model and extended to include the use of services provided by a practitioner and the self-directed or autonomous use of T&CM, which includes both products such as herbs and practices such as yoga.

Test and validation of the model
We conducted two studies to validate the model with the target population. First, an exploratory qualitative study with patients and health professionals in oncology to refine the variables. Then, we conducted a study with 200 patients. This study allowed us to find variables that were determinants of T&CM use in general. This study allowed us to find variables that were determinants of T&CM use in general and also allowed us to highlight the importance of psychological variables in use. This led us to use a valid questionnaire to include these variables, the ABCAM questionnaire, which was translated, tested, and validated in 50 patients.

Data collection and variables
The data were collected using a 54-item face-to-face questionnaire designed by the study team using the CAM healthcare conceptual model. Before their oncology consultation, patients were interrogated separately for 20-30 min in a waiting area.
To measure the prevalence of traditional herbal medicine use, we collected data about herbal medicine use in the last year (dichotomous dependent variable herb users/ herbal nonusers). We also distinguished users according to the purpose of use (cancer treatment, health promotion, or a combination of both purposes). A list of the most commonly used medicinal plants was used as support during the survey.
The independent variables were derived from the constructs of the conceptual model (the predisposing, enabling, and need variables).

Predisposing variables
These are the sociodemographic variables (gender, age, ethnicity (amazing or Arab), level of education, marital status, type, and size of family). Psychological variables include the perception of the severity of the disease and the effectiveness of the treatment and the emotions felt as a result of the disease. The Attitudes and Beliefs about Complementary and Alternative Medicine (ABCAM) instrument was used to measure attitudes and beliefs about T&CM. It is a 20-item questionnaire with 5 scales that measures the perceived efficacy and toxicity of T&CM and the influence of the decision to use these therapies (Mao et al. 2012). We first used principal component analysis (PCA) to analyze the validity of the scale and determine the factors and dimensions of this measurement scale. Among the 20 items of the scale that were analyzed, several variables were significantly correlated, the Kaiser-Meyer-Olkin index was qualified as excellent (KMO = 0.86), and Bartlett's test of sphericity was significant (P = 0.000). Therefore, the variables were factorable, and the model was valid.
Five latent variables were constructed by principal correspondence analysis (PCA): expected benefits of T&CM use, perceived scientific barriers (risk and toxicity of use), perceived socioeconomic barriers (geographic and financial accessibility), social norms of relatives (influence of family, friends, and patients), and social norms of caregivers (influence of physicians, nurses, and pharmacists).
Due to the fact that all Moroccans, regardless of their level of education, speak the same dialect (Darija), this approved and original English questionnaire has been translated into this language.

Enabling variables
They include occupation, monthly family income, environment, province of residency, medical coverage, and access to care in terms of distance and consultation time.

Need variables
They include both objective needs assessed from the examination of the medical file by an oncologist (type of cancer, stage of cancer, duration of the disease, treatment received, therapeutic strategies, number of treatment sessions, and comorbidities). The needs perceived by the patient (symptoms experienced by the patient and their intensity: pain, fatigue, insomnia, and anorexia).

Dependent variables
Two dependent variables were included: use of traditional herbal medicine (users and nonusers) and purpose of use: cancer treatment (CT), health promotion (HP), and combined treatment and promotion (CTP).

Statistical analysis
Univariate analysis of the data was used to determine participant characteristics and frequency of use, and bivariate analysis was used to identify determinants of use using Chisquare tests and the Fisher test for categorical variables. Student's t-test was used for quantitative variables.
Next, we tested the normality of the five ABCAM questionnaire variables (Kolmogorov-Smirnov and Shapiro-Wilk test). The distribution was not normal (P < 0.00); therefore, we used nonparametric Mann-Whitney U tests to compare the medians of the variables in users and nonusers of medicinal plants.
All analyses that showed statistical significance (p < 0.05) indicated that the variable was considered a determinant of herbal medicine use by cancer patients.
Sequential logistic regression modeling was conducted following the recommended Hosmer and Lemeshow steps to identify potential predictors of traditional herbal medicine use and the purposes of that use. Indeed, variables with p < 0.25 on the univariate analysis were inserted into the multivariate model to test for the influence of potential covariates. The final model included significant predictors (p < 0.05) (Hosmer and Lemeshow 2000). Odds ratios were also measured to determine associations between predictors and dependent variables.
We conducted sequential multinomial logistic regression to identify factors that could predict herbal medicine used for cancer treatment only (CT), health promotion only (HP), and combined treatment and promotion (CTP) to identify factors that could predict these three different purposes of the use of herbal medicine. Feasible factors with an alpha value below 0.1 were simultaneously added to the model, nonsignificant factors were phased out.
All measurements were performed via the Statistical Package for Social Sciences (SPSS) software (V23).

Characteristics of the participants
A total of 546 eligible cancer patients were invited to participate in the study. Of these, 530 participants gave informed consent with a response rate of over 90%. The mean age of the participants was 54.6 years (62.04 years for men and 51.68 years for women). The majority of participants were urban residents (57.5%, n = 304). The largest proportion of participants belonged to the Arab ethnic group (410; 77%) and the province of Beni Mellal (230; 43.4%). More than two-thirds of the participants were illiterate, and only 3% had a higher level of education. The majority of the participants had a low socioeconomic level since more than two-thirds had a monthly family income of fewer than 200 dollars (n = 363; 68.5%). Of these 530 patients, 40.2% (n = 213) of the women had a diagnosis of breast cancer, 35.7% (n = 189) had been admitted with advanced cancer (Stage IV) and 39% (n = 208) had received adjuvant chemotherapy after surgery. Prevalence, types, objectives, and patterns of traditional herbal medicine use The use of traditional herbal medicine during the last 12 months was found in 194 cancer patients in the Beni Mellal-Khenifra region. As shown in Table 1, the prevalence of this use was 36.6% and it varied with the type of cancer. The use of herbal medicine in order of importance was more important in the case of purely female cancers (breast cancer and cervical cancer), followed by liver cancer, ovarian cancer, colorectal cancer, endometrial cancer, and bladder cancer. Almost half of the users (n = 94) had the goal of permanently treating cancer, while 30% (n = 59) of patients aimed to enhance immunity, decrease cancer symptoms or decrease the effects of conventional therapy. A total of 21% (n = 41) of users had both objectives simultaneously.
Patients reported that their source of information was their family (n = 67; 34%), followed by patients they met during counseling sessions (n = 63; 32%) and the media (n = 32; 16%). Only 13% of users (n = 26) communicated the use of herbal medicine with health care providers, especially oncologists. The majority of nonusers of herbal medicine (n = 117; 35%) reported that the reason for nonuse was fear of adverse effects and lack of knowledge of the benefits of herbs. Of the nonusers, 13% (n = 45) intended to use herbal medicine in the future.

Determinants of traditional herbal medicine use
A bivariate analysis was conducted to identify variables associated with herbal use, in other words, determinants of herbal medicine use by cancer patients.

Predisposing factors
Sociodemographic determinants of use The demographic characteristics of medicinal plant use are given in Table 2. Women used more plants than men, with a highly significant difference (χ 2 = 15.87; p < 0.001). More than 40% of patients aged less than 40 years used medicinal plants. The mean age was slightly higher in nonusers, with no significant difference (χ 2 = 1.92; p > 0.05). No significant differences in herbal medicine use were found by other sociodemographic characteristics.

Psychological determinants CAM/MT attitudes and beliefs
Four ABCAM variables were strongly associated with herbal medicine use (Table 3). Indeed, herbal medicine users believed more in the positive results of T&CM and were highly influenced by close friends (patients, family, and friends) and perceived scientific and socioeconomic barriers to T&CM use. In contrast, the social norms of health care providers did not influence the use of this traditional therapy.
Perception of disease and conventional treatment Patients who perceived their disease as severe or very severe used herbal medicines more than patients who perceived cancer as not severe (between 34.3 and 56.5 vs. 17 and 44%) (χ 2 = 38.50; p < 0.000). The perceived emotional and social impact of the disease was also associated with this use (χ 2 = 39.20; χ 2 = 18.14; p < 0.000). In addition, the expression of anxiety and anger seemed to affect this use more, whereas patients who had reported feeling no emotion used herbal remedies less (52% versus 29%, respectively) (χ 2 = 25.27; p < 0.00). On the other hand, belief in the efficacy of conventional treatment was negatively associated with herbal use (χ 2 = 9.65; p < 0.05).

Enabling factors
The two groups of patients (users and nonusers) differed significantly according to monthly income; poor patients (with income less than 3000 DH, in this case, the minimum wage in Morocco for salaried workers) used medicinal plants more than other patients with a higher socioeconomic level (37.8% and 21.1%, respectively) (χ 2 = 7.26; p < 0.05). Patients who reported not having the ability to purchase medicines were more likely to use medicinal plants (χ 2 = 9.70; p < 0.01).
Patients who encountered problems at the hospital, including affordability and access to medicines, used herbs more than patients who found no problems in hospitals (χ 2 = 23.76; p < 0.000). In addition, payment for all or part of conventional anti-cancer drugs was associated with the use of traditional herbal medicine (χ 2 = 8.53; p < 0.01).
No differences in the proportion of herbal medicine use were observed concerning other socioeconomic characteristics and resources facilitating access to care, namely occupation, area of residence, province of residence, and distance to health facilities (Table 4).

Health determinants (need variables)
As shown in Table 5, the use of medicinal plants was strongly associated with the duration of the disease. The use of these remedies was lower before the first six months of the disease (21%). Then, it starts to increase to reach its peak between 1 year and 3 years (51%), with a very significant difference (χ 2 = 31.25; p < 0.000). However, the type of cancer was not associated with the use of plants, although women with breast cancer seemed to use plants more than other patients, and men with prostate cancer used them less, but without significant difference.
Patients with a less advanced stage of cancer (stage I) at admission used fewer medicinal plants (prevalence of less than 20%) than patients who were managed at the oncology center with a very advanced stage of cancer (stage IV), with a very significant difference (χ 2 = 11.62; p < 0.01). On the other hand, a patient with recurrent cancer used plants more (51%) than patients without recurrent cancer (35%), without any statistically significant difference (p > 0.05).
The presence and intensity of symptoms experienced by the patients were also strongly associated with this use. Indeed, patients who complained of intense pain, fatigue, or anorexia used more traditional therapies with a significant difference compared to patients who did not present these symptoms (p < 0.000). In addition, patients who perceived their health status as excellent used herbs less than other patients who saw their status as good, fair, or poor (χ 2 = 14.97; p < 0.01). Comorbid patients used herbal medicine more with a positive correlation between the number of diseases associated with cancer and the prevalence of use. Patients who had no other disease presented a prevalence of 33.6% versus 45% in the case of the presence of a single disease. This proportion reached 50% in the case of the association of two or more diseases with cancer (χ 2 = 6.52; p < 0.05), while the type of the associated disease did not affect this association.
The type of conventional therapy received was associated with this use. Indeed, patients who had received multiple therapies (chemotherapy, radiation, and surgery) tended to use herbal medicine (50%), whereas patients who received radiation therapy only used these herbal medications less (20%) (χ 2 = 16.47; p < 0.05). On the other hand, the therapeutic strategy did not influence this use (χ 2 = 13.80; p > 0.05), The therapeutic strategy is the complete protocol that the patient receives, and it differs according to the type and stage of cancer and also its biological and clinical health status. For example, a patient may receive all three chemotherapy, radiation, and surgery treatments, but the order of these treatments differs from one patient to another.
The use of several cycles of chemotherapy was associated with this use since those who received two or more cycles of chemotherapy were more likely to use plants (55%) than patients who received only one cycle of chemotherapy (37%) and patients who received no chemotherapy (24%) The distribution of expected benefits of T&CM is identical in both groups (users and nonusers) 0.000 0.000 The distribution of perceived socioeconomic barriers is identical for both groups (users and nonusers) 0.020 0.002 The distribution of perceived scientific barriers is identical for both groups (users and nonusers) 0.000 0.000 The distribution of social norms of relatives is identical in both groups (users and nonusers) 0.000 0.000 The distribution of social norms of care providers is identical in both groups (users and nonusers) 0.330 0.186 (χ 2 = 9.56; p < 0.01). On the other hand, the number of radiotherapy sessions did not affect the use of traditional herbal medicine (χ 2 = 2.32; p > 0.05). The management of cancer patients also affected the use of plants, since 45% of patients who were not satisfied with their management had used herbal medicine compared with 34% of users among patients satisfied with their cancer care and management (χ 2 = 3.92; p < 0.05). Regular follow-up of care and medical consultations was associated with the use of herbal medicine in Beni Mellal. In truth, patients who did not regularly follow their treatment and medical checkups tended to use these plants with a high prevalence of 70% (χ 2 = 8.75; p < 0.01). Among the health behaviors that were associated with the behavior of using medicinal plants, we found that the adoption of hygienic-dietary rules was positively associated with the use of herbal medicine (χ 2 = 5.82; p < 0.05). On the other hand, the exercise of physical activities was not associated with this use (Table 5). When a Chi-square test result is associated with more than one degree of freedom, we used the calculation of the residuals (raw, standardized and adjusted residuals)

Logistic regression to predict herbal medicine use in cancer patients
Each factor in the conceptual model was entered individually into the logistic regression model, with herbal medicine used as the dependent variable. Factors that had an alpha level of less than 0.25 were selected for entry into the final model using a sequential procedure (Hosmer and Lemeshow 2000). Table 6 shows the results of the logistic regression analysis with the predictor variables and their odds ratios, model coefficients, and significance levels. The model predicted 82.8% of the observations, with the model variables explaining 61% of the variance in herbal medicine use. The model is a good fit, as the Hosmer-Lemeshow goodness-of-fit test indicated a good fit for the model (Hosmer and Lemeshow 2000). For 530 observations. Pearson/2 (degree of freedom = 8) = 15.51, p = 0.05.

Predictors of objectives of use: results of multinomial regression
The predictive factors that were significantly related to the purpose of plant use are represented in Table 7, which presents the p-value, odds ratio, and purpose of use to which they were related. The expected benefits of traditional medicine were a significant predictor (p < 0.000) of the use of traditional herbal medicine for cancer treatment and health promotion (CTP). Indeed, those who believed in these positive outcomes were five times more likely to use traditional therapy for both reasons compared to those who did not.
In addition, cancer patients adopting the diet tended to use fewer herbal medicines to treat cancer only (CT) than patients who did not adopt the diet.

Prevalence and patterns of use
The results of this study indicated that more than one-third of cancer patients (36.6%) reported using herbal medicines at least once in the past 12 months. This finding corroborates the results of studies published in Morocco. Indeed, between 11 and 39% of cancer patients have reported the use of medicinal plants as complementary medicine. Nevertheless, this prevalence was slightly higher than the average prevalence recorded in Morocco between 2011 and 2020, which was 31.16% (Aboufaras et al. 2021).
These results are also similar to those found in the Middle East and North Africa (MENA) region. Indeed, 35.5% of Jordanian cancer patients used traditional botanical preparations (Afifi et al. 2010). In North America, the prevalence of herbal use observed among cancer patients is between 38 and 39% in the United States of America and Trinidad and Tobago, respectively (Richardson et al. 2000;Bahall 2017).
However, in Iran, another study found a very high prevalence of 84.1% of patients who used herbal medicine during chemotherapy sessions (Bazrafshani et al. 2019). This upward trend has been found recently in Europe as well. A study conducted in Germany (2021) showed a high

Predisposing factors
Sociodemographic factors Only gender was significantly related to use as a determinant and predictor among all the predisposing sociodemographic variables included in the conceptual model. Herbal medications were used by women more frequently than by men. These empirical findings confirm that women were more likely than males to utilize herbal medicine in the Moroccan cancer population, as previously documented by observational studies conducted in this population in Casablanca and Rabat between 2011 and 2020 (Kabbaj et al. 2012;Samouh et al. 2019).
The relationship between age and prevalence of medicinal plant use does not indicate a significant difference in use among different age categories. However, there was a slightly higher average age among the non-user group. The prevalence was relatively high among those under 40 years of age. The interpretation of the results of the age effect in cross-sectional surveys is completely confusing (Fouladbakhsh and Stommel 2007). As a result, contradictions in results have often been found. Contrary to what was observed, the age group between 41 and 60 years used traditional herbal medicine frequently compared to other age groups in Morocco (Kabbaj et al. 2012). Nevertheless, two other Moroccan studies found no association between the age of participants and the use of complementary medicines, where the predominant modality was herbal medicine (Brahmi et al. 2011;Rahou et al. 2017). A recent descriptive ethnobotanical study found a high frequency in patients aged 40-50 without a significant difference (el Ouahdani et al. 2021).
These results are contradictory to the literature on the use of herbal medicine, and natural products that seem to attract older cancer patients (over 60 years old), while young people tend to use mind/body and manipulation approaches (Abdallah et al. 2015).
Our study found no evidence of an association between education level and the use of herbal remedies. However, patients with low levels of education had a somewhat greater frequency. On the other hand, the usage of T&CM globally was associated with greater levels of education, which could be attributed to the fact that these studies did not isolate herbal medicine as a distinct treatment. For instance, the use of T&CM in America was strongly and independently predicted by the university education level. Higher levels of education enhance the propensity to use T&CM (Eisenberg et al. 1998;Molassiotis et al. 2005).
Regarding ethnicity, there was no ethnic difference in usage between Amazighs and Arabs in our study in Morocco. Furthermore, according to studies conducted globally, Caucasian cancer patients utilized T&CM at higher rates than African-American cancer patients; however, other studies, depending on the T&CM categories examined, found the converse to be true (Eisenberg et al. 1998;Fouladbakhsh et al. 2005).
In the Beni Mellal region and throughout all of Morocco, no other sociodemographic factors were connected to the use of herbal medicine. However, compared to married people, widower people with cancer were much less likely to utilize T&CM goods in other parts of the world (Fouladbakhsh et al. 2005;Fouladbakhsh 2006).

Psychological factors
Most of the psychological variables tested were associated with herbal medicine use in our study and strongly predicted this use.
More than half of the users of herbal preparations believed that traditional medicine could improve their health, reduce disease symptoms, and cure the disease. This was significantly associated with the prevalence of herbal medicine use and was an independent predictor of the medicine used in this study. On the other hand, respondents who reported that this medicine might have direct or indirect adverse effects used herbal medicine less.
In addition, the encouragement of patients by their relatives, belief in the competence of traditional medicine practitioners and the low cost of herbal products, belief in the seriousness of cancer, its psychosocial impact, and the ineffectiveness of conventional treatment were positively associated with herbal medicine use.
No national or international studies have used these variables to investigate their association with herbal medicine use in Morocco or outside Morocco (Neriman I˙nanc¸ et al. 2006;Afifi et al. 2010;Akyol and Öz 2011;Alishtayeh et al. 2016). This makes it difficult to discuss the results with other studies, especially in developing countries where the use is mostly traditional. Psychological factors have never been examined for this category of T&CM.
However, the ABCAM questionnaire used, which is based on Theory of Planned Behavior (TPB), and the constructs of this conceptual model will allow us to compare our results with the explanatory hypotheses of this model.
Our study confirmed the majority of the hypotheses of this model, except for the explanation of the behavior studied (use of CAM: herbal medicine) by social norms, notably those of the caregivers or care providers. Indeed, normally, high scores of expected benefits are positively associated with the use of complementary medicine. That is, the more positively patients perceive the outcomes of use, the more they use complementary medicine. Patients who use complementary medicine perceive fewer barriers than nonusers. Patients are more likely to use T&CM if it is recommended by family/friends and/or health care providers (Mao et al. 2012).
Thus, the social norms of the providers were not correlated with the social norms of the relatives as predicted by the model according to the PCA analysis we did, and this factor was not associated positively or negatively with the use of herbal medicines in our study. This result could be explained by the lack of communication between the caregiver and the patient, the lack of knowledge of health professionals on the subject, and the non-integration of complementary and supportive care in the new oncology centers in Morocco, especially since it is a traditional use that is not scientifically based.
Other studies using TPB have shown that perceived family expectations and medical team norms were associated with T&CM use. However, perceived control, including barriers, did not influence T&CM use (Hirai et al. 2008;ben Natan et al. 2016).
In addition, the other psycho-cognitive variables were also associated with the use of plants, such as belief in the severity of cancer, its psychosocial impact, expression of emotions such as anxiety, and ineffectiveness of conventional treatment.
Our results confirmed the importance of psychological factors in explaining and predicting the use of TMs/CAMs in general herbal medicine specifically.

Enabling variables
We found that herbal medicine use by cancer patients in Beni Mellal was highest among patients from low-income socioeconomic backgrounds. Indicators of this low socioeconomic level were monthly family income, inability to purchase drugs, financial inaccessibility to care, and direct payment for cancer treatment.
The results of the literature regarding the relationship between income, economic level, and utilization of T&CM have been contradictory. This may be due to the different categories of T&CM and cancers included in the studies. In addition, the measurement of income poses a problem of reliability. It is confidential data that cannot be verified, especially since in Morocco, many patients are reluctant to reveal their probably high income to continue to benefit from the indigent health insurance scheme; The results of our study were contradictory with the literature on T&CM use in Western countries, the majority of which indicates that higher income was a significant predictor of T&CM use in the cancer population (Eisenberg et al. 1998;Tas et al. 2005;Verhoef et al. 2005;Fouladbakhsh and Stommel 2007).
On the other hand, concerning the use of herbal medicine alone as a T&CM, our results are close to what has been observed on a national scale. Indeed, low-income Moroccan patients used herbal drug preparations more than cancer patients with a high economic level (Samouh et al. 2019). According to this study, patients with a low level of education and income are more sensitive to the use of traditional Moroccan medicine.
Regarding the enabling factors for the use of herbal medicine as complementary medicine, our results also agree with other international studies; monthly income less than expenses was a significant factor for this use in Nigeria (Ajah et al. 2019). In addition, in America, indigent patients use herbs more than other T&CM modalities (Richardson et al. 2000).
These contradictions could be due to the varying cost of T&CM modalities in the East and West. Modalities such as chiropractic, yoga, homeopathy, and even herbal medicines are more expensive than herbal preparations whose raw material is of low cost, and sometimes the plants are free when they grow spontaneously in nature after the population.
Our research confirmed the same results as previous studies. Indeed, the use of T&CM was not associated with occupation or with accessibility to conventional care (Richardson et al. 2000;Fouladbakhsh et al. 2005;Garland et al. 2013;Chin et al. 2020).
However, a study in Turkey discovered that living in an urban area was linked to the use of T&CM (Akyol and Öz 2011). However, since our study focused specifically on traditional herbal medicine and the poor use this form of treatment more frequently, it could be said that the results are more applicable to Western nations.
Furthermore, a Thai study that focused mostly on herbal medicine found that living in a rural area affected how often people used T&CM (Chukasemrat et al. 2021).

Need variables
This research confirmed that several health need variables of the T&CM healthcare model (perceived and assessed needs) were related to the use of medicinal plants in the Beni Mellal-Khenifra region. These variables were time since diagnosis, presence, and intensity of symptoms experienced by patients (pain, anorexia, insomnia, and fatigue), perceived health status, type of treatment received, number of chemotherapy cycles (each cycle is composed of several sessions (6 sessions of 3 or 4 h, for example)., regular follow-up of treatment, satisfaction with conventional management, number of coexisting conditions, and follow-up of an anti-cancer diet.
Our results, ethnopharmacological, therefore, suggest that herbal medicine was used by cancer patients whose health status was deteriorating because they might have some hope concerning conventional treatment. Health needs are insufficiently addressed in Moroccan ethnopharmacological studies as determinants or factors of herbal medicine use. This finding corroborates the existing international literature review, where the duration of cancer and multiple chemotherapies were independent predictors of T&CM use in Turkey (Tas et al. 2005). In Norway, the use of T&CM was also associated with the duration of the disease. Patients with cancer for more than 3 months used more T&CM (Kristoffersen et al. 2019). Another study found that the time since diagnosis exceeding 12 months was associated with the use of T&CM (Garland et al. 2013). In our study, these results were confirmed, as the prevalence of use was found to increase significantly between the first and third years after a cancer diagnosis.
The presence and severity of pain was a predictor of T&CM use, including the use of chiropractics, massage, and acupuncture practitioners (Eisenberg et al. 1998;Fouladbakhsh and Stommel 2007). A recent study in Norway found that the use of T&CM was associated with different symptoms presented by cancer patients attracted to it (Kristoffersen et al. 2019). A study specific to herbal medicine found that the presence of diarrhea or constipation was a predictor of use (Bazrafshani et al. 2019).
The coexistence of comorbid conditions could predict the use of herbal medicine in our target population, which confirms previous findings of research on T&CM use by cancer patients (Eisenberg et al. 1998;Fouladbakhsh and Stommel 2007;Bazrafshani et al. 2019).
The modalities of the conventional treatment received by the patient affect his use of T&CM; thus, the follow-up of chemotherapy treatment was associated with the use of T&CM, regardless of the adopted therapeutic strategy (single, multiple, palliative, or curative) (Richardson et al. 2000;Hyodo et al. 2005;Tas et al. 2005). This was found in our study as well since patients who received chemotherapy were more likely to use herbal medicine. In addition, the number of chemotherapy cycles was positively associated with this use. This also corroborates other studies that have found that the use of many of the conventional treatments was associated with the use of T&CM (Patterson et al. 2002). No Moroccan study has explored the association between cancer treatment modalities and herbal use.
In contrast, the likelihood of using medicinal plants does not seem to be affected by the diagnosis of the specific primary cancer, as is the case in our research and other studies (Fouladbakhsh et al. 2005;Bazrafshani et al. 2019;Mwaka et al. 2019).
Certainly, the prevalence was high in patients with breast cancer but without significant differences compared to other categories of cancer. Breast cancer has been found in several studies to be a determinant and predictor of T&CM uses (Tas et al. 2005;Alsharif and Mazanec 2019). A large study in Europe found that patients with pancreatic, liver, bone, and brain cancer used these complementary therapies much more than any other group of cancer patients (Molassiotis et al. 2005).
Concerning studies specific to herbal medicine, the association between the type of cancer and the use of plants has rarely been explored, even outside of Morocco. In fact, in Jordan, a single descriptive study also found that breast cancer was the predominant cancer among female users of complementary herbal remedies (50.6%), and lung cancer was the main cancer among male users of these therapies (18.1%) (Afifi et al. 2010). In Morocco, only one study found that patients with breast, uterine, colon, or leukemia cancer are the most frequent users of traditional medicine compared to other cancer patients (Samouh et al. 2019). Regarding the stage of cancer, Rahou and collaborators who looked at T&CM, in general, did not find an association between the stage of cancer and the use of these therapies by women with breast cancer (Rahou et al. 2017).
In contrast to the study by Rahou and colleagues in Rabat, the cancer stage was mentioned as a determinant and predictive factor for the use of herbal medicine internationally. Thus, patients with advanced metastatic disease use more herbal products and preparations than other patients (Poonthananiwatkul et al. 2015;Mwaka et al. 2019). The use of all T&CM modalities combined was associated with advanced cancer disease in several countries (Fouladbakhsh et al. 2005;Tas et al. 2005;Garland et al. 2013;Kristoffersen et al. 2019).
This variation in results from other studies is not only due to geographical reasons but also methodological; thus, the design of the questionnaire and the method of administering it could affect the results. It was found that the prevalence would be high if the questionnaire was administered face-toface compared to the telephone or self-administered questionnaire. In addition, administering the questionnaire by a trained person and presenting a list of drugs in the questionnaire could also increase the proportion of users (Horneber et al. 2012). On the other hand, not integrating T&CM into the clinical setting could negatively affect the rate of use because patients do not feel comfortable disclosing this use (Ernst and Cassileth 1998). Therefore, the prevalence of use at oncology centers is often underestimated (Eisenberg et al. 1998). In contrast, because we studied only herbal medicine, our results may reveal a higher prevalence than the results for the prevalence of herbal medicine if it is combined with other T&CM modalities (Asiimwe et al. 2020).
The analysis provided valuable information on the predictors of the use of herbal medicine by cancer patients for the first time in Morocco. Thus, the predictors of T&CM use in the study setting were gender, pain, emotional impact, social impact, expected benefits, perceived scientific barriers, perceived socioeconomic barriers, and social norms of relatives.

Study limitations
Most surveys are also subject to bias. For example, selection bias may result from the choice of the study sample. Measurement bias may also have occurred because some individual measures, particularly of some ABCAM questionnaire variables and subjective clinical variables, may be inaccurate.

Conclusion
The high prevalence found in the Beni Mellal oncology center should alert healthcare providers and public health planners to the importance of this use in cancer patients, the potential risks of this use, and the need to consider the factors behind this use. Identification of the user profile and predictors of this use would allow for targeted educational actions by clinicians so that these T&CM methods can be safely applied, particularly in the area of drug-plant interactions.