Determinants of practices for dengue diagnosis among healthcare professionals working in public hospitals of Abidjan , Cote d ’ Ivoire

1 University of Félix Houphouët Boigny, Abidjan, Côte d’Ivoire. 2 Epidemiological Surveillance Department, National Institute of Public Hygiene, Abidjan, Côte d’Ivoire. 3 Epidemiological Surveillance Department, National Institute of Public Hygiene, Bouaké, Cote d’Ivoire. 4 Université Alassane Ouattara, Bouaké, Abidjan, Cote d’Ivoire. 5 Vector Control Department, National Institute of Public Hygiene, Abidjan, Côte d’Ivoire.

Fredericks and Fernandez-Sesma, 2014) with four serotypes in circulation among human (Fredericks and Fernandez-Sesma, 2014;Messina et al., 2014).In Africa, the epidemiology and public health impact of dengue are not well understood or are poorly documented (Amarasinghe et al., 2011;Durand et al., 2000).In Cote d'Ivoire the first human infected case was reported in 1999 (Durand et al., 2000).Prior to that, dengue viral circulation has not been documented in Cote d'Ivoire since the early 1980s (Cordellier, 1984).
In 2008, two dengue cases were confirmed among travelers returning to their home countries from Cote d'Ivoire (a Japanese and a French tourists) (Durand et al., 2000;Moi et al., 2010;Ninove et al., 2009).The strengthening of the epidemiological surveillance led to the detection of a dengue outbreak in 2010 with 23 cases resulting in one death in Abidjan.These events confirmed the existence of the dengue virus in the country and its capacity to culminate in an outbreak.
However, in Cote d'Ivoire as in many African countries, dengue is likely to be overlooked and underreported because of a low awareness among health care providers, the pervasiveness of other febrile illnesses, and lack of diagnostic testing and systematic surveillance (Amarasinghe et al., 2011).This fact is conductive to antimalarial drugs misuse, therapeutic failures and subsequent malaria treatment resistance (Foumane et al., 2015;Shayo et al., 2015;Golassa et al., 2015).
In Cote d'Ivoire, it appears that there is a necessity to put in place a surveillance system for data gathering and the early detection of outbreaks.However, establishing epidemiological surveillance framework for dengue which may enhance its diagnosis and its prevention, required understanding the weakness and strength through capacity assessment, which is, based on the capacities knowledge, practices and behavior of healthcare workers in Abidjan and its suburbs.
The present study is aimed at identifying the determinants of good practices in the diagnosis of dengue among healthcare workers in Abidjan and its suburbs.

METHODS
This cross-sectional study was conducted from 3rd September, 2014 to 20th March, 2015 2015 in public health facilities of Abidjan and its suburbs.It assessed the knowledge, attitudes and practices of healthcare workers.Abidjan is the economic capital city of Cote d'Ivoire, with 4,707,000 inhabitants according the 2014 census (Institut National de la Statistique (INS), 2014).Abidjan has ten townships and two suburbs divided into ten health districts.
The study population was composed of 400 health professionals who worked within the services of general medicine, infectious disease, pediatric and medical emergencies in the public sanitation network of Abidjan and its suburbs.The sample size was calculated with Epitable software version 6.0 (CDC Atlanta) by taking into account the following parameters: 1. Expected prevalence of good practice: 37% (Dhimal et al., 2014) 2. Required accuracy: 5%; 3. Confidence level: 95%; 4. Non-response rate: 10% The calculated sample size was 397 healthcare workers.A simple allocation scheme was applied.40 health professionals were interviewed in each health district.
A proportional allocation method was applied for the number of staff from each occupation to take into account sample's representability.Hence, the designed sample included 120 doctors (30%) and 280 nurses (70%).The healthcare facilities were randomly investigated.
The dependent variable was the practices referred to dengue diagnostic among healthcare professionals.Good practice was defined as the capacity of healthcare workers to make differential diagnoses of fever cases, to demand blood samples for laboratory analyses, to research primary or secondary cases in the immediate environment of non-malarial patients, and to identify these cases as a dengue-like disease.The independent variables were sociodemographic status, occupation, knowledge, and attitudes toward dengue.
The data were collected by face to face interviews with a questionnaire developed using the World Health Organization (WHO) guidelines on dengue disease (World Health Organization, WHO, 2009), the Integrated Diseases Surveillance and Response guidelines (World Health Organization, WHO, 2010) and the studies published by Dhimal et al. (2014) in Nepal and Shuaib et al. (2010) in Jamaica.Quantitative questionnaire were administered to each interviewee.
The data analysis was performed with R software version 3.1.2(2014-10-31).The knowledge and practices variables were scored according to the number and weight of related items included in the questionnaire.Thus, good practices and good knowledge required a minimum score of 60% and worse practices and worse knowledge were scored under 60%.The threshold of 60% to scale knowledge and practices about dengue disease was determined by consensual agreement from national expects in Public health and epidemiology.
With regard to "Attitude", two variables were identified to describe it.Yes/no questions were applied to collect healthcare workers' impressions about their perception of the seriousness of the disease and their perception on the fact that Cote d'Ivoire can be at risk of dengue.
The relationship between the dependent variable and the explicative variables was determined by multiple logistic regressions with stepwise selection.All of the variables with a significance level less than 0.2 (Freitas et al., 2012) in the bivariate analysis were selected for multiple regression.Knowledge was defined as a main variable.The final model contained variables with a significance level of less than or equal to 0.05 after adjustment on other variables.Variables that had a matching strength were kept less than or equal to 0.05.The robustness of the final model was tested by a Hosmer-Lemeshow test with a significance threshold of 0.05.

Ethical considerations
Before the study, the objectives and the methodology of the study was sent in writing to each health facility directorate to obtain their consent.The anonymity of the subjects' data was ensured by using deidentification codes.The data were compiled, and the results presented so as to avoid identification of individual through their answers.
This study did not include any treatments or unpleasant procedures.If a participant refused to answer, there were no negative consequences.There was no risk associated with participation in this study.This was voluntary, and participants did not receive remuneration.Each participant was permitted to end the interview at time without reason and without fear of persecution.

Description of the study sample
Out of the 400 health professionals, 340 were effectively investigated, with a participation rate of 85%.The participation rate was 97% among doctors and 80% among nurses.Four doctors (3 males and 1 female) and 37 nurses (20 males and 17 females) abstained from answering the questions, either arguing that their participation had not been approved by their superiors or they were very busy.Nineteen nurses were not investigated because of their work schedule and limited number of staff.
The median age of investigated healthcare workers was 37 years old with an interquartile range of 33 to 44.The age ranged from 23 years old (minimum) to 65 years old (maximum).The respondents were composed of 199 males and 141 females, with a sex ratio (M/F) of 1.41.The median number of years of experience was 6 with an interquartile range of 2 to 14 years (Table 1).Of the 340 respondents, 46 were specialists (14%), of whom 41 (89%) were medical doctors.56% of the specialists were noted to be pediatricians.
The distribution of healthcare workers by type of health facility showed that 55% of them came from reference facilities, namely general hospitals (GH) and university hospital centers (UHC) (Table 1).

Knowledge, attitudes and practices (KAP)
Seventy (21%) health professionals showed good knowledge score (Table 1).Regarding attitudes, 253 (74%) health professionals knew that dengue was a serious illness (96% doctors and 66% nurses).There were 146 health workers who perceived the risk of dengue outbreak in Cote d'Ivoire.In fact, 59% of nurses and 53% of doctors asserted that Cote d'Ivoire was at risk of a dengue outbreak.From a practice's standpoint, 21% of the participants provided answers linked to good practices.

Relationships between the dependent variable and independent variables
It was observed that gender was significantly associated with practices; males had better practices compared with females [OR = 2.07; CI 95% = (1.20 -3.70)] (Table 2).Additionally, variables measuring attitude, such as perception of disease seriousness [OR = 2.61 (1.34 -5.64)] and the perception of the risk of dengue outbreak in Cote d'Ivoire [OR = 3.16 (1.86 -5.48)], were significantly linked to practices.
From the perception of knowledge, health professionals in Abidjan and suburbs who had good knowledge of dengue fever, had also good practices for dengue diagnosis [OR = 2.73; CI 95% = (1.53-4.85)].

Multivariate logistic models design
Table 3 presents the final results of the logistic regression after stepwise selection.After adjustment on other variables, the health professionals who had good knowledge about dengue had also good practices for dengue diagnosis [aOR = 2.69; CI 95% = (1.37 -5.32)].
Ceteris paribus, male healthcare workers had better

DISCUSSION
This study was performed to sustain the establishment of a specific surveillance system for dengue.Since the early 2000s, Cote d'Ivoire has declared several dengue outbreaks located in Abidjan and suburbs.Unfortunately, the outbreaks were discovered a posteriori.Therefore, it appeared necessary to assess the knowledge and practices of dengue fever among healthcare professionals in Abidjan and its suburbs.The results from this study led us to establish the following findings.

Demographic and professional characteristics
Of the 340 healthcare professionals interviewed, 34% were doctors whereas the majority was nurses (66%).Ho et al. (2013) evaluated a sample in which 51% were doctors and 49% were nurses.Dubé et al. (2011) assessed a sample in which 57% were doctors and 43% were nurses.In Cote d'Ivoire, medical doctors represent 30% of healthcare professionals (SIDA, 2012).Therefore, this study was designed to have representative data from both healthcare occupations.The distribution by age highlighted that 50% of the health professionals interviewed in Abidjan and its suburbs aged less than 37 years, ranging from 23 to 65 years old.The study population was relatively younger than that of Lee et al. (2011), who performed a study in which 69% of medical doctors were 41 years old. Lee et al. (2011) evaluated a sample with age distribution similar to Tan et al. (2009), who performed a KAP study in Taiwan among doctors on communicable diseases in the private sector.The study found that 83% of the surveyed doctors were more than 40 years old.Jain et al. (2015) found that their sample of healthcare professionals was relatively younger than that of the present study.The difference between the formally cited and the study finding can be ascribed to the fact that this study took into account nurses and medical doctors simultaneously.
The distribution by gender showed that male healthcare professionals were significantly older than female healthcare professionals (p < 0.0001).Similarly, doctors were significantly older than nurses (p = 0.023).This finding could be explained by the fact that doctors enter the workforce at a relatively higher age, have a longer university curriculum (eight years of study vs. three years of study) and receive their pension later than nurses (65 years vs. 57 years).
The study found a sex ratio (M/F) of 1.11 among nurses, that is, 53% were men while 47% were women.Among medical doctors, the sex ratio (M/F) was 2.31, with 70% men and 30% women.Ho et al. (2013) found far more males (87%) than females among medical doctors, whereas among nurses, female represented an overwhelming majority (97%).
The results from Ho et al. (2013) study in Taiwan suggest that among healthcare providers there may have been an association between occupation and gender.This result was also corroborated by Tan et al. (2009) study which was conducted in Taiwan among private doctors.Tan et al. (2009) found in their sample that males were predominant (87%).However, the results from Dubé et al. ( 2011)'s study in Quebec highlighted a predominance of females in medical professions.
The study found that 31% of health professionals had more than 10 years of experience.In the study published by Ho et al. (2013), 51% of surveyed healthcare professionals had more than 10 years of experience.The respondents in this study also had fewer years of experience than those investigated by Dubé et al. (2011) in Quebec.In fact, 65% of health professionals investigated by Dubé et al. (2011) had at least 15 years of experience.
Three sources were identified from which participants acquired information on dengue.Shuaib et al. (2010) and Dhimal et al. (2014) identified televisions and radios as major sources of information on dengue fever, followed by school.However, the study found that trainings (50%) were the main source of dengue information, followed by other form of press (30%).Furthermore, 25% of the respondents reported getting information on dengue via radio and television.The difference between the study results and those of Shuaib et al. (2010) andDhimal et al. (2014) might be due to the fact that they conducted surveys on the general population in Jamaica (Shuaib et al., 2010) andNepal (Dhimal et al., 2014), respectively.

Practice referred to dengue diagnosis among healthcare workers
It was found that good knowledge of dengue was associated with good practices for dengue diagnosis [adjusted OR (aOR) = 2.69; 95% CI: (1.37 to 5.32); p = 0.0041].Study conducted in general population in Nepal about dengue's knowledge, attitude and practices found a positive correlation between knowledge and practices (Dhimal et al., 2014) A study conducted by Ntambwe on multidrug-resistant tuberculosis among healthcare workers in Maseru (Malangu and Adebanjo, 2015) achieved the same result.This result leads to the belief that education plays an important role in practices enhancement (Merga and Alemayehu, 2015;Saaka, 2014).However in this study, medical doctors had worse practices than nurses [aOR = 0.37; 95% CI: (0.18 to 0.70); p = 0.0030].This confirms Ho et al. (2013)'s findings in Taiwan.Unlike Ho et al. (2013), Shuaib et al. (2010), found that education cannot always be correlated with knowledge.The explanation of this difference in this study can be found in the fact that in Cote d'Ivoire, nurses are more involved in the epidemic-prone diseases surveillance than medical doctors.Males from this study were found to have significantly better practices than females [aOR = 1.88; 95% CI: (1.05 to 3.46); p = 0.0357].In Tawain, Tan et al. (2009) reported similar findings but with no significant correlation.The perception that Cote d'Ivoire can be at risk of dengue epidemic was positively associated with good practices [aOR = 2.59; 95% CI: (1.47 to 4.62); p = 0.0011].Some studies conducted on other epidemic diseases such as influenza demonstrated that the perception to be at risk of a threat strengthen the practices (Lau et al., 2010;Hollmeyer et al., 2009).

LIMITATIONS
The investigation used a sampling method that allotted an equal number of participants for each investigated district.This may have induced an non-response rate because some districts had lack of healthcare professionals, leading the study to have a possible limiting ability to detect certain associations or an overestimation of certain relationships.However, the selection of institutions was essentially randomly-based, and the 85% participation rate may have limited potential bias.

Conclusion
Dengue fever remains a misunderstood disease in sub-Saharan Africa, particularly in Cote d'Ivoire.The study was mainly descriptive in nature and the design allowed the identification of healthcare professionals from Abidjan and its suburbs who were able to diagnose dengue as part of the surveillance of epidemic-prone diseases.The results of the multiple logistic regression model have established that the practices of healthcare professionals in Abidjan and its suburbs are linked to occupation.Therefore, nurses had better practices for dengue diagnosis than medical doctors.This study shows that there is a need to train healthcare workers in Cote d'Ivoire in terms of diagnostic capacities for fever diseases.Also for the better accomplishment of this screening, in all level of health facilities some rapid tests must be applied to triage patients.

Table 1 .
Description of the study sample (knowledge, attitude and practice study on dengue among healthcare professionals).

Table 2 .
Univariate logistic regression (KAP study on dengue among healthcare workers in Abidjan and its suburbs).
*, Only variables with a p-value ≤ 0.2.The variables with a p-value > 0.2 were health zone, years of professional experience and specialty.**Do not know.

Table 3 .
Multiple logistic regression final model (KAP study on dengue among healthcare professionals in Abidjan).