Things to bear in mind when developing an internet intervention

Things to bear in mind when developing an internet-based intervention

This chapter is for people who are new to creating Internet-delivered interventions. It is beyond the scope of these help files to provide a detailed guide on how to do this, but we have listed below some key tips and issues to consider.

Our useful resources and links page includes helpful web links for finding out more about important things to bear in mind when developing online interventions.

Tips for designing a web-based intervention

Intervention designers often base their web-based interventions on interventions that have proved successful in other formats (e.g. face-to-face or printed delivery). But the way in which web-based interventions need to be presented is quite different, for several reasons:

1. Internet users are demanding – they are impatient and are used to good-looking, entertaining, easy to use websites. People will not just sit and read pages of text on the computer – to get the information across you will need to condense it, using plenty of bullet points, boxes and optional links to further information. A fantastic guide to creating a website that it is engaging and easy to follow is Steve Krug’s book Don’t make me think!

It helps to break up the text with pictures and you can also use more engaging formats such as quizzes, games and stories. Unless you are very creative it is a good idea to get input from graphics designers, illustrators, or multi-media designers – the most professional interventions have even employed film directors for their videos!

2. Internet users like being in control. You may think your end-users should log on, complete 6 long questionnaires and then follow a set course of sessions for 8 weeks – but they are used to browsing the web anonymously to choose what they will look at whenever they want to. Web users are typically unwilling to log on or give their email address or other details, and like to be able to click on what they want to do and see. Highly motivated and committed end-users (e.g. people who have signed up for an internet course of cognitive-behavioural therapy for a serious problem) are more willing to log on and follow a recommended programme.

3. Computers can’t read. If you want your intervention to give advice or feedback based on what end-users enter then you need to convert what they enter into a numerical or categorical variable.

Collecting numerical/categorical information from end-users

• If you ask the user to type in a description of what physical activity they did today or what their activity goal is then you cannot give feedback based on this text (though you can get the computer to record it and present it later – perhaps if you want to remind people what goal they set).

• If you ask the user to select an amount of physical activity they want to achieve as their goal from a drop-down list of 5 options, and then type the number of minutes they did light, moderate or intensive activity, you can use the software to give feedback about whether they are meeting their goal.

Learning from other interventions designers’ experience

There is no need to try to reinvent the wheel – you can learn from others who have created website interventions. It really helps to have someone on the team (or at least providing advice) who has done it before.

You can get good ideas for the format and components from looking at other web-based interventions. The demos and interventions on the LifeGuide Community website can also show you how to create similar components using LifeGuide.

Learning from your target end-users

You will very quickly learn that no-one can provide you with better advice about website development than your target end-users! Website designers may advise you about the best look and structure for your site and topic experts may insist on particular ways of delivering the intervention – but it is vital to find out how real end-users feel as early as you can, as this is often quite different from what was expected.

Input from end-users can be obtained by:

• including one or two members of your target population on the design team
• holding focus groups at key points in the development
• creating a panel of end-users who are happy to give you feedback on repeated occasions (it may help to ask them to try out the intervention at home and keep a diary of it)
• using the ‘think aloud’ method to observe people actually using the intervention, and get their immediate reactions. Include all the types of end-users you want to reach, e.g. young and old, experienced and novice computer users, people with disabilities etc.

An advantage of using LifeGuide is that you can keep testing and modifying it as you develop it based on how end-users react. You can also collect details of how the end-user navigates through the site and what pages they use most often using the data collected in the intervention manager of LifeGuide.

Planning your intervention development

Planning for different types of advice

You will need to create a different version of advice and feedback for each sub-group created by tailoring (i.e. personalising advice based on the user’s inputted characteristics).

• If you just tailor by gender and whether the user is young or old then you will only need to create 4 versions of the advice (2X2, e.g. for young women, older women, young men and older men).

• If you also want to also give different advice depending on whether they are improving or deteriorating, whether they are low or high in motivation, and whether they are low or high in confidence, you need to produce 32 different combinations of advice (2X2X2X2X2).

Planning your time

The length of time it will take to develop your intervention obviously depends on the complexity and quality you need to achieve, and whether you have already developed your materials. For example, making high quality illustrative videos can be expensive and time-consuming. So it pays to think carefully about whether features that you would like to see in the intervention will really affect engagement and outcomes for your end-users.

Most interventions have four stages: planning (working out how many sessions you will have and what they will contain, emails you will send, and so on), writing content, building the intervention (creating pages and writing logic) and testing the intervention.

Assuming that the basic intervention materials have already been developed (for example for face-to-face delivery), then as a rough guide to the time taken for one person working full-time to create the webpages and logic using LifeGuide:

• a short twenty-item questionnaire with basic logic commands may only take a few days
• a detailed session for a CBT intervention with 20 pages linked by complex logic commands may take a month (double this for the first one you create, but by the time you are doing the last one in a series it may take less than 3 weeks)

On top of this, you will also need to allow time for:

• preparation – e.g. bringing in a designer for the look and feel of the site, consulting users about what components are needed

• creating the logic for running an automated evaluation – it can take two months to set up complex procedures for screening, randomisation, reminder follow-up emails etc.

• testing the logic for every possible end-user of a site (e.g. does an 80 year old male who has recently been in hospital go to the advice that matches his situation) and testing that the site works as it is supposed to (e.g. do links and buttons take you to the pages that they are meant to).

• more testing – we cannot stress enough the importance of testing. It would be really disheartening if you finished developing your intervention and then found that it didn’t do what you wanted it to do!! So, before you put the same piece of logic into every session of your intervention, you might like to test it is doing what you expect it to do the first time you add something new. Regularly previewing your intervention is also highly recommended.

• qualitative piloting with a range of end-users throughout (including gaining ethics approval for pilot studies)

Prioritising your intervention features

Since it is difficult to be sure how long development will take (but it always takes longer than you expect!) it is useful to prioritise the functions you wish to include in your intervention. You can use the MoSCoW approach to identify functions you want:

• ‘Must have’ (essential)
• ‘Should have’ (important, but can be worked around if necessary)
• ‘Could have’ (nice to have but not critical)
• ‘Would like’ (for future projects).

Creating a trustworthy and accessible intervention

It is your responsibility to create an intervention that is – and is seen as - accessible and trustworthy. End-users like to know who created a website, what your credentials are, how you can be contacted, what evidence it is based on and when it was updated. The Health on the Net Foundation provides guidelines for good practice and will inspect and certificate health-related websites on request.

It is difficult to make a website equally accessible to everyone (for example, different formats increase accessibility for people with different types of impairment), so it is important to research what features will make it easiest for your target population.

If you are going to evaluate your intervention, then you will need to obtain ethics approval in the usual way. You should, of course, gain prior informed consent to collect and analyse any data from end-users. There is a growing field of literature that you can consult on ethical issues relating to web-based research. You also need to bear in mind legal issues, especially if you are giving advice that could potentially cause harm.

Security

Researchers are naturally concerned about security, especially when it comes to the data they collect during a trial. All LifeGuide servers are secure servers (https), which are hosted by the University of Southampton and behind the firewall of the department of Electronics and Computer Science. We have followed the standard techniques for Web system security; for instance researchers' passwords are not stored in plain text in the database and LifeGuide uses the security standards implemented by Java's Spring Security.

While we have taken all reasonable security measures in the design of LifeGuide, it is important to note that the biggest cause of security breaches is through weak passwords or sharing of passwords. It is important to have a standard operating procedure in place that ensures that end-users create strong passwords and don't exchange them with each other.

References

Krug, S. Don’t make me think! A common sense approach to web usability. New Riders Publishing, USA, 2006.