The results presented in this paper are based on the total experiences from 9 case studies. The individual results are not presented, as the objective of this paper is to argue for the relevance of the video observation technique to provide insight into sociomateriality and by doing so inform and improve health information technology development. Further, the aim is to share our experience on the most important lessons learned on benefits and challenges when using the video observation technique within healthcare settings – and not to present the results of the individual cases, which have quite different objectives (e.g. assessment of clinical benefits and labor-savings when implementing HIT).
During our studies, we have developed a generic guideline on how to conduct a video observation study:
1. contact to the management at hospital level (explain objective and methods)
2. contact to the management at ward level (explain objective and methods)
3. preliminary visits to wards (information to clinicians and other relevant actors, studies of logistic and organizational issues by personal observations and interviews)
4. video observation
5. data-analysis and interpretation
6. data validation with clinicians and other relevant actors
7. data presentation
Besides, we have gained thorough experiences on where, when and how to use video observation to inform and to improve HIT development. Based on our experiences, video observation provides benefits in studies with the following objectives:
1. to inform and improve the design of new health information technologies through studies of specific clinical work practices in delimited clinical settings
2. to study changes of specific clinical work practices before and after the implementation of new health information technologies in delimited clinical settings
3. to identify potentials for new ways to organize clinical work practice - including potential labor savings - when implementing new health information technologies through studies of specific clinical work practices in delimited clinical settings
4. to document current clinical work practice for future research purposes (e.g. before and after new HIS is implemented) through studies of specific present clinical work practices in delimited clinical settings
A number of benefits are common to the study objectives listed above - as well as a number of lessons learned on how to manoeuver as a researcher collecting data by use of video observation within healthcare settings.
During our studies, we have found that video observation is most beneficial for studies of specific work practice within delimited clinical settings. Trying to study a wide range of clinical work practice at a number of different clinical settings at once is extremely time consuming and should only be done if based on thorough considerations on why and how to conduct the study.
Used within the healthcare sector, video observation permits us to explore context dependent
clinical sociomaterial work practice, often involving staff with different professions in contrast to in usability laboratories or in clinical set-ups. In the latter unforeseen disruptions and communication challenges can be hard to imitate no matter how realistic the clinical setting has been set up [36
In a video observation study on the impact of the Electronic Health Record (EHR) on sharing information’s between nurses at morning meetings, the nurses sought information’s in two different EHR systems, while at the same time making handwritten notes at other schemes and papers. The interactions between the nurses and the different artifacts went on extremely fast. When analyzing the video data, new insights were gained every time, we revisited the field through the sequences of recorded data [44
]. Thus, the video observation technique is capable of capturing real time and continuous activities and hence of providing data, that allow us to study interactions between clinicians and technology in local time and place as well as over a period of time. Video observation provides the basis for insight, understanding and interpretation of the complexity of clinical work practice, as the complexity is recorded and analysis can be broken into smaller sequences and re-visited over and over again. These rich details cannot be achieved using other ethnographic observation methods. Besides, in contrast to other ethnographic methods – including personal observation - data from video observation allow us time and again to revisit the observation site and gain new insights, alone or together with the clinicians, with other stakeholders or with researchers, without having to physically return to the field.
In the same study as mentioned above [44
], also the impact of the EHR on the exchange of information’s during ward-rounds was studied. At the wards, a number of work tasks went on simultaneously: communication between the patient and the clinician’s and between the clinician’s themselves, decisions on new medication and treatments, different clinical measurements etc. When subsequently analyzing the data, we recognized that our original focus on information flow was too narrow, because the EHR turned out to have a major impact on other aspects also, e.g. the organization of work. Thus, video observation has the advantage compared to personal observation and hand written notes that the recordings can be revisited time and again presenting revised research questions. Despite that the focus for the observation (by hand or video) is subjectively decided by the observer before or when the observation takes place, the recorded data are rich and embrace more than the initial focus.
When analyzing the data from an explorative study on medical secretaries work practice before the implementation of the EHR together with the involved secretaries, it became obvious that during interviews and personal observations conducted previously, only part of their work practice had been captured, partly because of the complexity, partly because of the routine and tacit nature of their work practice [16
]. Thus, compared to data from personal observations, video data become a data repository allowing both contextual knowledge and the analysis process to be revisited and shared – and validated - with the involved clinical staff. As to validate the data generating process, the data repository permits the clinicians to access and discuss if the data recordings actually do represent their clinical work practice and further, if it represents the work practice that we, as researchers want to explore according to the study objective. Letting the clinicians themselves validate the video data, is a most reliable way of validating, e.g. the camera position in a clinical setting, because when a certain position is selected, others, which might also have an important meaning regarding the study objective, are left out.
Additionally, studying the video recordings in collaboration with the clinicians – and other relevant professionals in HIT development – provides an excellent basis for a dialog about understanding clinical work practice.
Overall, by providing us a thorough insight into the complexity of sociomaterial clinical work practice, the common benefits on video observation is insight into and a better understanding of the work practice of health care professionals.
Below we will present our most important lessons learned during three different phases in HIT- development:
the planning phase
the data collection phase
the data analysis and interpretation phase
We consider the planning phase the most important of all phases, because a well-planned study is crucial to the success of the following phases. Therefore, in this paper more attention is given to the planning phase compared to the other phases.
The planning phase
The first and most important step in any
study – no matter the methods and the techniques used - is the preparation of the study methodology, as this constitutes the “roadmap” from start to end [55
]. However, a precondition for formulating the study methodology is clearly formulated objectives, as this decides the content of the methodology (e.g. the theoretical approach, the design, the data collection methods) [52
]. Compared to other ethnographic data collection techniques, the video observation technique generates large amounts of data in a short period of time. It is therefore important to know exactly: why, where, whom, when and how to conduct the study, ie. to have a precisely formulated objective – and methodology. The fact that in “real life”, time, economic conditions, the study settings etc. very often are obstructive for an optimal methodology is not an excuse for the researchers not to focus on clear formulated objectives.
When conducting research – e.g. evaluation studies - in collaboration with external contracting authorities, the objective is often defined by the client beforehand. According to our experiences, such study objectives are often rather “loosely” formulated. An example is: “to provide an assessment of the clinical benefits of X system”. This objective might seem straightforward at first, but if not elaborated, a number of questions will inevitably arise later on in the study-process:
what is meant by “clinical benefits”?
from who’s point of view should the benefits be assessed? (clinician’s, management, patients - or other relevant actors in HIT development?)
by which indicators should clinical benefits be measured?
Thus, it is necessary to clarify the study objective to avoid any future misunderstandings and - what is also important - to adjust the study objective to meet “real life” constraints when it comes to economic conditions, time, study settings etc. However, a precondition for the researchers to be able to clarify a study objective is a thorough insight into the clinical context in which the study is going to take place. In a hospital setting, this will include:
number of wards
number of sections in each ward
number of employees – and professional groups to be involved in the study
the shifts (in a 24/7 setting) relevant to study
Some of this information can be achieved through homepages, literature and dialog/interviews with relevant key informants, but – based on our experiences - preliminary visits to the study locations to gain the needed insights through personal observation and interviews are mandatory. Based on the knowledge achieved through site visits etc., the researchers must then reach an agreement on clear and measurable objectives to prevent major problems and misunderstandings later on in the process. First then, the additional elements of the methodology (e.g. design and data collecting methods) should be formulated.
When the objective indicates that the video observation technique is beneficial, it is important to reflect and decide on the kind of video observation (Figure ). However, this can often be read from the (clearly and precisely) formulated objectives, as theformulation of these often (directly or indirectly) tells, if the researcher should interact with clinical practice or not.
The next step in the planning phase is to decide on:
1. the overall perspective: specific work practices, certain professional groups or…?
2. the focus: e.g. which specific clinical work practice performed by which professionals at which wards at which time by which researchers?
3. which camera angels should be used: e.g. should the video be fixed or roving - and does the setting (e.g. space, patients, work-procedures) allow a camera to be placed in the optimal position?
4. what is left out by the decisions taken in 1, 2 and 3 - and how does this impact the results?
5. how many researchers will do the recordings?
6. for how long time (hours, days, weeks) to record?
The list above is not exhaustive, but it comprises the issues found most important during our studies. Common to all questions are that the answers are closely linked to the study objectives. Thus, when answering the questions, the study objective must simultaneously be readdressed.
An important step during the planning phase – which should be taken as soon as the study objective is formulated - is getting permission to do video recordings within the clinical setting, ie. at the wards. Our experience is that a well prepared study and clear objectives are imperative for achieving the trust and confidentiality of the management necessary to gain permission to do video recordings within a ward.
However, having gained permission from the management does not imply that we walk straight into e.g. a hospital ward and start recording. To establish trust and willingness to participate in the study among the clinical staff is an important part of the planning phase. Thus, at preliminary visits at we inform the clinicians on the study objective, what our presence mean to their daily work, what is expected of them, etc. This is an issue, which is also stressed by Heath et al. [47
]. During our video observation studies, we have only met few clinicians who have declined to be video taped, and we have met none, who - after having been well informed about the study objectives and methods – have insisted on not participating. However, prior to data collection, informed consent should be obtained from all staff members who are to be observed cf. the section on ethical considerations.
When the planning phase is completed, the next steps in the study process should be far less time consuming – given that the study methodology is well prepared and all the precautions mentioned in this section have been taken.
The data collection phase
As mentioned in the section on ethical considerations, a very important consideration, when video-recording within a clinical setting, is how to avoid video recording the patient – and especially the patient’s face. When recording within a ward, the optimum solution to this is to place the camera at the headboard of the bed. From this angle, it is possible to capture most activities going on in the room without recording the patient’s face. It can be difficult to avoid recordings of patients, when e.g. following a clinician up and down the corridors with a roving camera while he/she is performing a ward-round. If a patient’s face by accident is video-recorded, we delete that sequence.
During recordings, it is important fully to adhere to the methodology formulated in the planning phase. This means that when the observation method decided on is e.g. non-participant observation (fly at the wall), the researcher should refrain from interfering with the clinicians and the clinical work practice. If not, the data-validity is compromised, as the data gathered will be influenced by the researcher’s interaction. If the video observation method is participant observation, the researcher may ask the clinicians to reflect on their own practice (fly in the eye). This may provide a clarification and elaboration of socio-material activities, issues and situations while video recording, and thereby provide more in depth and elaborated information of e.g. reasons behind specific clinical work practice. Information, that otherwise would remain tacit knowledge and/or stickyinformation [58
It is our experience that when clinical staff is video recorded, they often try to perform certain tasks according to “what the books tell”, instead of how they themselves have appropriated work practice to fit the clinical context and the resources available. Thus, our presence does
affect the way they behave when performing their daily clinical tasks. However, the more and the better the clinicians have been informed on the study objective beforehand, the better we have been able to establish contact and trust - which is a precondition for diminishing this bias. Thus, after a while (often only one-two hours), they seem to be less affected by our presence. This view is supported by Nøhr et al. [19
]. We have found that it is important to dress up like the people we observe, e.g. in white coats and to use small video-recorders and to hold them in the least possible eye-catching position in order to attract as little attention as possibly (Figure ).
Figure 2 Position of the researcher and the camera. Figure shows one the authors (Anna Marie Høstgaard) dressed as the clinicians and the size and position of the camera. The example is from the evaluation of the GEPKA project .
The data analysis and interpretation phase
Analyzing video-recorded observation data differs from analyzing data collected through other ethnographic methods by the huge quantity of data – both visual and audial. If the study has not
been conducted from a fixed, clear and precise methodology, problems often arise during this phase, because of the researchers loosing focus and missing the “red thread” when navigating through the (high time-consuming) analysis phase. Part of - or in worse case - all the collected data can then prove to be useless, because they prove irrelevant to the study objective. During one of our studies, the objective was revised by external contracting authorities half way through the study. Originally, the objective was to assess the clinical benefits of the fully implemented system from the clinician’s perspective. This was changed into an assessment of the clinical benefits of the implementation of only the first part of the system. This meant that a substantial part of our baseline data became irrelevant [44
If, on the contrary, the data have been collected from a well-planned methodology, the researchers still have a large amount of data – but they also have an overall plan for the analysis phase. We have developed a systematic method including three major steps on how to manage this phase:
Step 1: Create an overview of all the data (often many hours of recordings) in order to identify the sequences relevant according to the study objectives. Look through all the video recordings and at the same time outline the work practice activities in a chart: what is going on: when: which technologies are used: who is using them: and where does the activity take place. Figure shows an example of a work practice analysis of a physician’s ward-round.
Figure 3 Chart providing information on activities in a ward office. Figure shows a chart providing information on activities in a ward office, when a physician and a nurse are preparing for the ward round: at what time (time-line to the left): (more ...)
The chart allows us to identify which data sequences are the most relevant according to the study objective, as a large quantity of the collected data always prove not to be important (e.g. walking down the corridors, waiting for the results of tests, the clinicians having breaks).
Step 2: The data-sequences identified as relevant according to the study objective are then analyzed in depth. This includes transcribing and mapping the clinicians – as well as other relevant actors – actions. We will not go into details with this process, as other researchers have already done this [10
Step 3: The last major step is to interpret the results from step 2. According to our experiences, clinical work practices are very hard to compare across contexts, settings and time (e.g. before and after implementation of HIT), because of the individual nature of clinicians’ work. Besides, also the patients involved are different. Therefore comparing clinical work practice should be done with great caution. We share this view with other researchers [44
When interpreting video-data – as any qualitative data – it is very important for the researcher to be explicit and transparent about the study paradigm as this shows the researchers ontological position. As mentioned in the method section, the same data can be interpreted differently depending of the ontological position held by the researcher.