Development of the system
A Wireless Health Outcomes Monitoring System (WHOMS) prototype was designed and developed in order to satisfy two main objectives:
1. To allow patients to receive and self-report structured questionnaires via either WAP [25, 26] or the Web [27];
2. To allow the physician to examine data reported on questionnaires through a graphical and chromatic interface.
These objectives can be met in the following ways (Figure 1):
• (Scenario 1) Periodical sending of questionnaires to patients with mobile handsets. The questionnaire shipment uses a WAP/GPRS connection to send a "WAP Push Service Indication" message to the patient's mobile phone. The patient can see the questionnaire on the phone display via the GPRS connection.
• (Scenario 2) Questionnaire (10 symptoms questions in the prototype) completed by the patient. Using their mobile phone, the patient completes the questionnaire following the directions presented on the display. Questions are displayed one at a time and a set of answers is presented in a menu (Figure 2). The patient chooses the most appropriate answer according to their symptoms. Alternatively, for patients who prefer to use personal computers, a reserved online area can be accessed, where questionnaires can be compiled or overall reports containing previous answers can be displayed (note this internet option for patients was not trialled in the present study).
• (Scenario 3) Answer management. Using a reserved online area, the physician can examine patient's symptoms according to their questionnaire answers. The graphical and chromatic representation allows the doctor a quick and clear vision of how the patient's symptoms are evolving. A light flashes by the names of those patients that present seriously modified symptoms, so that the physician can identify the most significant changes at once and immediately take the necessary action. The parameters that determine the type of change that will cause a flashing signal can be customised by the physician and are specific for each questionnaire (Figures 3 and 4).
Functional architecture
The functional requirements have been translated into a system architecture made up of the following modules (Figure 5):
Identification system: this component manages accounts, profiles assigned to system users and related identification. Once identified, the user is authenticated and authorised to use certain functions based on their profile and role recognised by the system. The authentication process is based on the user's MSISDN (mobile number) if connected through WAP, or on the user's account (login and password) if connected through the internet. The users' authentication guarantees both transmitted and accessed data privacy.
Questionnaire management system: this represents the heart of the system, and allows us to manage questionnaires (i.e. create, modify and delete). Through this module, it is possible to assign a questionnaire to one or more patients by specifying the examination elapse and the delivery recurrence. Questionnaire results are processed using the analysis function of this module.
Messaging system: this supplies the interface for sending SMS and MMS messages in order to solicit patients for recurrent compilation of questionnaires.
Rendering system: this module controls the questionnaire display based on user's type of handset. As far as possible, it aims to make the display not dependent on the type of device used.
Technological infrastructure
The prototype has been developed through use of open source software. The data structure was implemented through the MySql database [28], while the application logic was based on PHP language [29] and the Apache Web Server [30]. A Nokia emulator [31] was employed to develop and run preliminary tests on the WAP component of the system. The WHOMS was designed to offer a tool that can comprehensively enter each user's home. The channels used at present are WAP and WEB (Figure 6).
For most patients, mobile phones are easily available at reasonable prices, which is why we have recruited GPRS technology for questionnaires. In contrast, the use of ad hoc media requires a bigger investment in terms of time and cost, and would have brought about many organisational issues linked to its propagation and maintenance, in addition to user training.
In order to limit problems linked to the display of questionnaires on handsets, only one mobile phone model was used in the present study. This allowed prompt implementation of a prototype and rapid feedback from pilot users. The mobile phone selected can display questionnaires with a very simple, immediate and appealing graphical interface.
A demonstration has been developed on the internet that allows us to test some of the functions offered by the system. It is possible to simulate a questionnaire completed by a patient, and then see the monitoring report for the physician updated with this new information transmitted by the patient http://www.qlmed.org/whoms/.
While the ultimate aim is for patients to also use this system via the internet, the current study focussed on patients accessing the system by mobile phone.
Patients
In the present study, the WHOMS was tested using a sample of 97 cancer inpatients. The patients were asked to complete a ten-item questionnaire using a mobile phone.
The survey was conducted in 12 sample days during two months (January–February 2004) in 5 Hospital's Units at the Istituto Nazionale Tumori of Milan. All inpatients (with the exception of those in the immediate post-surgery period or with visible physical impairments or slipping) were invited to use the system after receiving a ten minute explanation and demonstration of how the questionnaire was to be completed. The aim of this pilot study was to determine whether this mobile phone-based WHOMS questionnaire method would be successfully used by patients. Thus, patients would not directly benefit medically from this particular questionnaire compilation as their doctors would not be invited to see the answers or make clinical decisions based upon them. For this reason patients were volunteers rather than compulsory recruits. All patients used the same model of mobile phone – a Nokia 6600 – and the GPRS connection was via Vodafone.
Data regarding gender, age, years of education, primary tumor site, surgical unit, use of communication technologies (calls or SMS from mobile phone, internet or email use by personal computer), time to compile the questionnaire and the number of items missing were also collected. Patient's free observations after compilation were also documented.
The questionnaire
For this pilot study we used an easy questionnaire regarding 10 symptoms (pain, lack of energy, worry, weight loss, cough, difficulty sleeping, shortness of breath, problems urinating, lack of appetite, difficulty concentrating) extracted from the MSAS-SF [17]. Patients were requested to select the response that best described the extent to which the symptoms distressed or bothered them during the past week. They were asked to select from one of the following responses: Not at all / A little bit / Somewhat / Quite a bit / Very much.
Statistical analysis
The association between the outcome variables (ability to compile and acceptability of WHOMS) and patients' characteristics was examined firstly at univariate and then multivariate level using logistic regression models. Regression diagnostics and indices of model fit were applied to evaluate how well the models fitted to the data. Results are presented in terms of odds ratios and their 95% confidence intervals (CIs).
In order to reduce the dimensionality in the set of predictors, and to avoid sparseness of data, the four variables dealing with use of communication technology (mobile phone calls, mobile phone SMS, internet, email) were collapsed into a single score representing the number of communication tools they declared to have used in the week prior to hospital admission. Using this approach, the Familiarity with Communication Technology (FCT) score obtained ranged from 0 (no familiarity – no tool used in the week before) to 4 (high familiarity – all the tools used in the week before).