Longitudinal Research – Expectation versus reality
Andreea is in the 3rd year of her PhD (ASBS) running a qualitative study on food consumption and community food growing. Emma is in the 2nd year of her PhD (ICAMS), working on a study assessing cognition and mood after a stroke. We met on the ‘Grad on the Island’ residential course and through hearing each other’s talks realised we were both running longitudinal studies.
Longitudinal studies are observational in nature, and collect data on the same individuals over a certain period of time. Although our studies are quite different in terms of topic, scale and setting, there were a lot of similarities between our experiences. Andreaa has now finished data collection, whereas this is ongoing in Emma’s study. Here we both reflect and discuss the realities of conducting a longitudinal study and share some insights into the challenges.
Designing and establishing the study
We both agreed that at the early stage of designing your research study, it’s important to think carefully about the time scale necessary to answer your research question. Having a longer time scale means multiple follow-ups, which would then involve more time, more money, and a higher likelihood that people might drop out. So, it’s important to think carefully, and ensure that you have a rationale for your frequency of follow-ups. It’s common for researchers who are just starting out to overestimate the number of participants to involve in their studies. study. Make sure you carry out your sample size calculations, and remember to always take into account people withdrawing - you may have to recruit more to compensate for this.
Andreea: This was a very important part of my research as I was setting up the study and conducting the data collection by myself. I aimed to slightly over recruit to get my necessary sample of 15 - I based this number on the previous literature on my topic. Initially I recruited 23 people and was left with 15 for my longitudinal data (so it worked out well!) My study took place over three years, and my research design involved interviewing participants three times throughout the course of 18 months and doing four observations with them, as well as a participant completed diary. My participants were satisfied with the required time commitments, and I reassured them throughout my study that their contribution and time is very valuable to me.
Emma: The data for my PhD comes from a larger clinical research study funded by the Stroke Association called APPLE – assessing post-stroke psychology longitudinal evaluation. For this type of study, a much larger sample size is needed (our initial target was 500 participants). However, what seems plausible on paper is very different in reality and we soon realised that it would be a challenge to reach this number from one hospital alone. Various hospitals across Scotland help recruit to the study to help us get sufficient numbers. We currently have 230 participants and will stop recruiting in November this year. APPLE involves many follow-ups; after the initial assessment, I see the research participants again at 1, 6, 12 and 18 months. We were aware that this would involve a lot of work to carry out but necessary for us to understand changes, short and longer term, after a person’s stroke.
Recruit and Retain – Making your research sound appealing
Andreea: This was the most challenging aspect of my research and definitely something I underestimated. Recruitment took a lot of time; time researching and contacting organisations that were suitable, time visiting them and meeting with gardeners, coordinators or community organisers. Unfortunately some of this time invested was lost as people often never showed up for the recruitment interviews. This was something I struggled with a lot during my first year so the best advice I can give you is: don’t take it personally! It can be overwhelming to dwell on why some people dropped out over the year; could I have done more? Was it something I did? It really isn’t you. I realised recruitment for a longitudinal study is an ongoing negotiation.I tried to make sure that it was as easy as possible for my participants to take part. This involved being flexible with times and locations for the interview, which adds another element of field work ethics and safety, and I wrote about that in a previous blog post. Interviewing the same people over a period of time is different from a single point of contact study – like one interview. It’s a balancing act between formality and familiarity; you develop a more complex relationship than just the formal research dynamic as you get to know your participants. For that, I think it is important that you check in regularly with your participants to maintain that connection and remind them you’re still there.
Emma: I work in an NHS setting and ‘recruit’ patients who have had a stroke as my research participants. Taking part in the study can be quite time consuming and this may put people off and so I highlight at the beginning that all follow-ups are not compulsory. To optimise recruitment and more importantly retain your participants you essentially need to ‘sell’ your research and think about any incentives for taking part. For me that means explaining why this area of research is really important, and how taking part in research will help our understanding and the care of future stroke patients. There is no direct benefit of taking part in the study (there is no intervention) but as with all clinical research patients receive more attention and oversight of their care, for example in our study if we are concerned about the patient’s cognition or mood we will refer them onto the clinical team. Patients should also never be out of pocket for taking part in research so if you can cover travel expenses it is very helpful.
Attrition is a common problem in longitudinal studies. It is inevitable that some of your participants will drop out – so don’t take it personally!
Andreea: Drop out was the second biggest issue for after initial recruitment. I’ve recently finished my data collection and I did not realise just how anxious I had been about people dropping out until the very last minute. Like I mentioned earlier, I recruited 23 participants and only 15 remained for the duration of my study. My participants were quite varied in terms of age so there were no immediately obvious factors that might influence drop out rates other than people’s commitment. When drop outs happened, it was often that people stopped replying to my messages. At that point you have to decide when you’ve sent enough follow up messages, and when to give up on that participant. This can be hard but it’s ok to move on! The silent “ghosting” can be daunting but again remember it’s not your fault. For me, I found that when the participants stopped responding they also stopped going to the gardens so I took that as an indicator that they were not that interested in food growing overall which is useful for my analysis.
Emma: The big issue around drop-out is that you are potentially left with a biased sample – are the people who complete the study really representative of the population of interest? About 20% of my research participants have withdrawn so far. Clearly there are factors for withdrawal which are outside of our control. Working with a patient population and generally older adults means unfortunately people’s health will deteriorate over a follow up period of 18 months so a common reason of drop-out is due to ill health or death. Some patients also feel like they have too many hospital appointments so doing research on top of this is too much. Some flexibility in data collection however can help with the retention of participants. Thankfully we have a budget that covers the cost of taxis to bring patients to the hospital, or for us to go see the patient at home. This has been really helpful, as many of the patients have limited mobility. We also have an alternative assessment, which we deliver over the telephone if face-to-face is not possible. Thinking about alternatives such as these and costing them in at the beginning can make a big difference so spend some time thinking about how you can minimize drop-out rates. Also, consider how you will deal with missing data when it comes to your analysis.
Scheduling and time-keeping
Andreea: We both reflected on how important time keeping is. Even for a small-scale study like mine I still relied on a spread-sheet I designed to follow my data collection stages and keep my supervisors updated on my progress. Longitudinal studies take up a lot of your research time so scheduling is a good way to make sure you’re making the most of your time and your participants’ time as well.
Emma: I work on a larger scale, currently following up 230 research participants and continually recruiting more. This requires a lot of organisation and planning. I work as part of larger research team of PhD students and research nurses who also carry out assessments and we have a secure spreadsheet where we track all of the follow-ups. If, like me, you will be working in a similar set up, communication and teamwork are key to make this work. Some advice for organising follow-ups would also be to not underestimate how long it takes to organise these visits; some people will be hard to get in contact with, especially if they’re working full time, so it is important to plan ahead if you need to get data within a certain time window.
Alternatives to getting your data
Andreea: For me, there were no obvious alternatives to the longitudinal data I was collecting but my backup plan was to collect single interview data to supplement my research. Another approach could be analysing someone else’s data. This is becoming more common as the ESRC for examples requires researchers to make their data available to other researchers. There are archived data sources on UK Data Service.
Emma: If in the scenario you do not get enough data, you should be thinking whether there are any existing databases to complement your own data collection. In my area of research for example there are many databases out there from different research studies that you can apply to use, for example the UK biobank. Applications are required and some databases will also have costs involved these so speak to your supervisor first.
Emma: Longitudinal studies are demanding and bring many challenges but allow us to answer important questions; in healthcare for example they are useful for assessing aetiology and prognosis. On a personal level I find it enjoyable because you get to know your research participants well.
Andreea: Overall, I did not know what to expect when I started my longitudinal study – I definitely feel I was quite naïve about the technicality and the challenges it involved so I would advise anyone who is just starting to be prepared!
What are your experiences of longitudinal research? Or what would you like to know about the practicalities of doing it? Get in touch in the comments.