Profile and preferences users of doctors consultation application in Indonesia

Researchers in healthcare predict the need for healthcare applications will grow rapidly. In Indonesia, doctor consultation applications began to emerge, but research on doctor consultation application still a few. The purpose of this study was to identify the user profile and preferences for doctor consultation application in Indonesia. The data were collected using an online questionnaire via Google Form that consisting of 3 parts: user profiles, user preferences, and doctor consultation application. There was 129 response and only 111 can be used. SPSS version 24 was used to test validity and reliability, descriptive data, cross tabulation, multi-response calculation, independent t-test. Consensus calculation used to find the preference of doctor consultation application in Indonesia. The results show that there are 6 most important factors that developers of doctor consultation application should consider. The most popular applications are Alodokter and Halodoc where Halodoc is a leader in doctors consulting application, so developers can make these two applications as a reference. The developer can target the market using user profile from this research. In addition, the features that should be in the application are the features of consultation (chat), disease information, purchase and delivery of medicine, and consultation (call).


Introduction
The modern world is rapidly changing human behavior every day. An ever-expanding technology can transform the conventional way of human life in a more modern direction. The development of increasingly sophisticated technology can affect daily human activities thus create new challenges and opportunities in every aspect of human life including health care.
The growing number of smartphones every year supports an increasing number of applications every year, especially mobile applications. It is predicted in 2019 will be 2.5 billion smartphone users worldwide [1]. This is certainly a business opportunity for mobile application developers because the market size is large and predicted become larger. Information technology in the healthcare industry has taken advantage of electronic devices for service and also health monitoring of its users. Wallace et al believe that mobile access to information made possible by mobile computing devices has the potential to change how medicine is learned and practiced now and in the future [2]. Ventola's predicted an uptrend regarding the use of mobile devices and apps in health care and will be widely incorporated into nearly every aspect of clinical practice in the future [3]. Pai wrote that about 46 percent of healthcare professionals want to introduce smartphone apps into their practice within the next five years, according to a survey conducted by market research company Research Now of 500

Methodology
The research in this area still new and from literature study, the study of Thinnukool about mobile healthcare in Thailand is suitable with this research because of their research in the Asia Pacific and corresponding with the purpose of this research. The main tools in this research are questionnaire which is based on Thinnukool et al research, with adaptation to fit the Indonesian users.
There are four dimensions to evaluate doctor consultation applications: user opinions on doctor consultation application as a tool in the initial medication, user opinions on Graphical User Interface (GUI), user opinions when using doctor consultation applications, and user opinion of the reliance on the use of application functionality [6]. The questionnaire for this research consists of three parts: user profile, user preferences, and a general question about doctor consultation application. Before the questionnaire distributed, the first step is face validity of the whole item of the question by asking several people if there are any ambiguous questions. After refining the questionnaire based on the face validity, the questionnaire distributed via Google Form. The usable questionnaire only for people who have downloaded, use doctor consultation application and passes the filter question. The next step is the validity test, followed by a reliability test. Data processing consists of descriptive data, crosstabulation, and multiresponse calculation of questionnaire part 1 and 3. The next step is processing the part 2 with independent t-test and consensus to calculate user preferences of the apps. After data processing, the next step is to analyze the data. The final step is the conclusion and direction for further research.

Data collection
The questionnaire has three parts where parts 1 is demographic of the respondent to capture user profile, parts 2 is user preferences in doctor consultation application, and part 3 asking a general question about doctor consultation application. The first part of this questionnaire contains a demographic profile of respondents that asking gender, age, educational background, employment status, income, family status, residence, types of mobile device used (IOS or Android) and questions about doctor consultation application that respondent use. Part 2 in the questionnaire use Thinnukooll et al research with adaptation. This sections are measured using Likert scale 1 -5, which is 1 = Very Unimportant, 2 = Unimportant, 3 = Neutral, 4 = Important, 5 = Very Important. The filler question is used in this part to segregate between a response that can be used and remove in the next step. The Item in questionnaire Use of doctor consultation application as a tool for initial medication

A1
The apps providing information on the initial medication and resulting in an effective treatment

A2
The apps can provide accurate information on the healthcare as same as professional advice from pharmacists and doctors A3 Using the apps can make personal healthcare better A4 Using the apps can change behaviors and encourage users to care more about their personal health. A5 The apps information is safe to user health A6 Using the apps can reduce the bill for medication when the users get sick A7 Using the apps can facilitate the users in the initial medication Graphical User Interface (GUI) B1 Buttons on the screen are easy to find B2 The screen is clearly distinguishing between foreground and background B3 Screen colors are appropriate B4 Each of the menu images represents understandable functionality B5 The number of menu is appropriate to use B6 Menu list is easy to use B7 Text sizes are appropriate B8 Contrast of the display is beautifully balanced B9 Overall user interface design on the screen Application Usage C1 The apps is easy to use C2 Every feature available on the app works well C3 The apps stable runs on the mobile device platform Reliance on the apps functionality D1 User confidence in the apps regarding instruction in care or treatment D2 User belief that the apps can provide accurate information about medicines D3 User confidence that the information provided by the apps can help the user to understand their healthcare D4 User confidence that the information provided by the apps will not harm the user D5 User confidence that the apps useful as health support It is intended to get users of doctor consultation application as respondents, so the voluntary response sample was conducted with the distribution of questionnaire using Google Form link in social media such as Line, Whatsapp, and Telegram and got 129 responses. After checking the responses through the filter question, there are 111 questionnaires that can be used. The next step is to validate the questionnaire using Pearson correlation and all item is valid because the total r score calculation is bigger than r from the table. Reliability test using Cronbach's alpha shows that the questionnaire is excellent reliability with 0.9 of Cronbach's alpha value.

Results
The first part of the questionnaire that capture user profile is presented in Table 2 and the third part of the questionnaire presented in Table 3. In the first part of the questionnaire, the cross-tabulation 4 analysis is divided into 9 sections by linking gender, frequently used applications with several other variables including age, education, occupation, income, status, residence, mobile devices, applications used and duration of application. Cross-tabulation results by linking gender, frequently used applications with some other variables including age, education, occupation, income, status, residence, type of mobile devices, applications used and duration of application, it can be deduced that most users of the consultation application doctors have the following relation: The third part of the questionnaire has three question that can choose more than one answer, so the multi-response used to process this. The three question is what doctor consultation application that you have ever used, for who do you use the apps, and what are the important features in the apps, coded as M1, M2, and M3 respectively and presented in Table 4 to 6. Results from multi-response are: 1. Halodoc is the highest percentage of applications that user use. 2. Most of the user use the apps for themselves. 3. Features of consultation (chat), disease information, purchase and delivery of drugs, consultation (call) are the most interesting feature for doctor consultation applications.
The second part of the questionnaire was processing separately using an independent t-test. This test used to test whether there is the difference between male and female in response the question because this research wants to know the preference for both male and female, so if there is any difference, then the question will be drop in the next steps. The null hypothesis for an independent ttest is that the two population have equal means [7]. The results of the independent t-test shown in Table 7.        Table 7 shows that there are two items that have significance smaller than 0.05, meaning that A1 and A4 reject H 0 . It shows that population male and female have a different opinion on item A1 and A4. The 22 other items have significance more than 0.05, meaning that the two population have the same opinion on the items.
The next data processing is a consensus which is used to get the preference of the users of the doctor consultation application. Consensus has two common meanings, one is a general agreement among the members of a given group or community; the other is as a theory and practice of getting such agreements [8]. Likert scale is the most widely used and controversial because it is ordinal scale but converted to numbers and treated as interval data. Research from Gail and Norman shows that a parametric test can be used to analyze Likert scale responses [9,10]. Tastle and Wierman introduce a mathematical measure that permits a logical determination of dispersion around a category value to determine a group consensus from Likert Scale data, which called consensus [8]. This is the rules that must be satisfied before any measure can be considered a viable solution to the Likert scale consensus problem [8]: 1. For a given (even) number of individuals participating in a discussion on some question of interest, if an equal number of individuals, n/2, separate themselves into two disjoint groups, each centered on the strongly disagree and strongly agree on categories, the group is considered to have no consensus. 2. If all the participants classify themselves in the same category of the Likert scale, regardless of the category, then the consensus of the group is considered to be complete at 100%. 3. If the mix of participants is such that n/2 + 1 participants assign themselves to any one category, the degree of consensus must be greater than 0, for the balance in the group is no longer equal at the extreme categories. Hence, a complete lack of consensus generate a value of 0, and a complete consensus of opinion yields a value of 1 and every other combination of Likert scale categories must result in a value within the unit interval [8]. The equation to calculate consensus shows in equation 1 where  x is mean of X, d x is the width of X (d x = X max -X min ), pi is the probability (relative frequency) of outcome X i (which ranges from 1 to 5), i is an index, and n is the number of categories.
This is an example of the item in question A5 and the consensus summary depict in Table 8 with descending order from larger to smaller.

Conclusion and direction for further research
Doctor consultation application developer should target the market based on the user profile to a female with age under 35 years old, with education background under bachelor degree, having a job as employee or students, income under 9 million rupiah monthly, single status, who uses apps for themselves and domiciled in urban areas. They also can use Halodoc as a benchmark because of the highest percentage of users and focus on features that interesting for users such as chat and telephone call consultation, disease information, purchase and delivery of medicine.
Consensus gives information on the preference of factor that most important for users and developer of doctor consultation apps can focus on several first factors. For example, from the first quartile or 6 factor, the most important factors for users are user confidence that the information provided by the apps can help user to understand their health care, menu list is easy to use, contrast of the display is beautifully balanced, user confidence that the information provided by the apps will not harm the user, user confidence that the apps useful as health support, each of the menu images represents understandable.
Further research should replicate this research and took a larger sample since this is the first research conducted, to make the research more generalized.

Acknowledgment
Thank you to University of Pelita Harapan for the support of this research.