Reference Concept Maps

Just a little update in terms of words, but a picture is worth a thousand words ;-)


A big issue in the healthcare literature is the different terminology and the definition that belongs to a specific term. To tackle this problem in a scientific manner the decision is made to search the literature, find and sort the definitions that are provided in the articles and make a reference concept map per term. A reference concept map takes into account how many articles uses a specific term in their definition and you can make a flowchart of this as I did in the following pictures:

Every box contains a term that is used by the articles to describe an electronic health record (EHR) and electronic medical record (EMR) respectively. These terms are grouped in 7 categories, visible in bold. The blue ovals contains the term that is used the most in its category. The red lines and green boxes show which definition is most used by the scientific literature. The little table visible on the EHR picture is less relevant for now.

Renewed scope and selected research method

As stated in my last post I had a meeting with Jo Bollen, a highly experienced consultant in the field of Electronic Health Records, on March 5. After a very informative and lively conversation we came to the conclusion that the best way to conduct a scientific research would be to focus on the eight teaching hospitals (in Dutch: "academische ziekenhuizen") in The Netherlands. These eight teaching hospitals has to select their EHR system through an strict European procurement which consists of huge questionnaires, takes a lot of (costly) time and therefore cost a lot of money. This European procurement can either be a closed or an open procurement, but I will focus on the closed procurement.


In case of the closed European procurement there are only five companies which participate in becoming the supplier of the EHR system for that particular teaching hospital. In random order these companies are: SAP/Siemens, Chipsoft, iSoft, EPIC, and McKesson. Of the eight teaching hospitals already four has completed their closed European procurement in the last five years. Two combined their procurement and two other did it on their own, in total there are three closed European procurement finished. Surprisingly, all three closed European procurement gave different suppliers of the EHR system that is the best to implement in the teaching hospital. This is at least a bit strange. How can it be that the outcome of a procurement by one teaching hospital differs from another procurement, while teaching hospitals doesn't differ that much from each other??


After consulting with my mentor we will focus on the latter question and therefore set the scope to the eight teaching hospitals in The Netherlands for phase 1 of my master thesis. Why only for phase 1 you may ask? Well, this is because of the fact that for a good master thesis you'll have to broaden your scope in the end and see what the results of phase 1, so the results of the teaching hospitals, means for the selection of EHR in general hospitals, which will be phase 2 of my master thesis. The partners with which I will cooperate during phase 1 will be obviously Jo Bollen and hopefully another great partner, but I will wait to mention them until it is certain. Also for the second phase I hopefully have a great partner, but I will not mention them until it is certain.


So that is my plan for my master thesis, but before I can start with the master thesis I will have to finish the literature study. Below you can read my research method if you are interested. 

In the field of Electronic Health Records a lot of literature is available, a simple search for ‘electronic health records’ provides more than 1.8 million articles in Google Scholar. Obviously not all articles can be read for this literature study. Therefore an approach is determined to go through all the available literature in a structured manner.

To make sure that the approach would be thoroughly and properly a small literature study is done by hand searching to determine what would be the best way to conduct the literature study. This hand search led to the following approach where the literature study has been divided into two phases (Holly, Salmond, & Saimbert, 2011):

      -        Phase 1 where the articles would be ranked by the number of citations in Google Scholar;

      -        Phase 2 where a citation chase would be conducted on the top ranked articles to find missing

            articles.

As there is a lot of available literature and the author is relatively new to the subject, phase 1 is a start where the authenticity of the authors is the key factor.  Phase 2 has to follow as there could be a few main articles which most of the top ranked articles cite that is important to this subject. This way the massive amount of articles has been brought back.

To determine how many articles should be included in phase 1: top-3 ranked, top-5 ranked, top-10 ranked, or even higher, there has been made use of the Hirsch contemporary h-index. The original h-index is a way to perform fair ranking of scientists as it accounts for both productivity and impact (Sidiropoulos, Katsaros, & Manolopoulos, 2007). A drawback of the original h-index is its inability to differentiate between active and inactive scientists (Sidiropoulos et al., 2007). (Sidiropoulos et al., 2007) states that their contemporary h-index, which gives more weight to recent articles and thus rewards academics who maintain a steady level of activity, solves this problem.

The program which is used to generate the ranking in number of citations is called Publish or Perish (Harzing, 2007). Het program automatically counts en ranks the number of citations an article has received in Google Scholar. The program generates the contemporary h-index per search query as output and this is used to determine how many articles should be included in the short list. Unfortunately, when a search query provided a contemporary h-index of 10 (e.g. search query = “definition of ehr"), not all of the 10 articles are relevant. To overcome this problem another step is included in phase 1, which is to verify whether all articles that should be included according to the contemporary h-index are relevant.


Please feel free to provide feedback or a comment, it will be highly appreciated. Thanks.

Structure literature review

Hello everyone, just a quick update about my work so far. Last week was the famous and infamous week of Carnaval in The Netherlands and therefore not much work has been done after the last time. Just before the Carnaval I had a meeting with my supervisor, Pieter Van Gorp, in which we decided that the best way to conduct a literature review is to start with a quick literature review about how other literature reviews have been done. I have scheduled 8 hours for this to make sure that I don't spend to much time on it.
The upside is that I will conduct a literature review in a scientific manner, the downside is that the work that I have done so far is, of course not wasted or superfluous, but a bit of a pity.

To tackle the problem that many scientific papers state different definitions of important terms we probably are going to make use of a technique called Reference Concept Maps. This technique provides a way to objectively state, out of multiple definitions, one single definition of a term.

Please feel free to leave any highly appreciated comments or ideas.

Update literature review

After my last post on January 30^th I had one exam of a master course left on February 2^nd. The good news is that I passed all my master courses in the last semester, but unfortunately these master courses took all my time so I couldn’t work on my literature review. Monday February 6^th was the first official day that I could start working fulltime on my literature review. The first week was a bit messy, which way I want to go, how I can get there, which literature is applicable and (maybe more important) which don’t!

This week is more productive as I made use of the protocol by Vanwersch. It is a protocol which provides more structure to a literature review and I could definitely use that.

The protocol consists of different steps and through these steps I hope to deliver a structured literature study:

1)      Search for an initial set of studies by making use of search keys.

2)      Through relevance screening and quality screening the initial set of studies is brought back to those studies which are relevant and of good quality. The screening procedure is done by implementing inclusion and exclusion criteria.

3)      Extract useful data from each included study.

4)      Summarize and compare the extracted data critically.

These search keys are ideally derived from the research question(s). My research question(s) are not final yet, but my broad subject is: ‘Provide a structural approach in the selection process for Electronic Health Records using Feature Modeling’.

In this broad subject are a few important factors like ‘Electronic Health Records’, ´Feature Modeling’, and ‘selection process’. I’m not sure if I will keep these factors, but for now I think they are sufficient. I have a meeting with Pieter Van Gorp later today in which these factors will be discussed.

I think the best way to tackle the literature review is to go through all four steps by Vanwersch for one search key before I start with the next search key. 

The factors listed above are not suitable as search key (e.g. the search term ‘Electronic Health Records’ provides 26,700 studies!!). I decided to start with the factor ‘Electronic Health Records’ as this is a correct factor for sure. But how could I bring that enormous number of studies back to a manageable number? First of all I decided to do a little brainstorm session (see attached picture). What do I want to know from this factor and which sub factors come into play?

The first search keys I used are:

·         “definition of electronic health records” (18 studies found, 13 useful)

·         “definition of EHR" (58 studies found, 22 useful)

For step one, the initial set of studies, I entered the search key into Google Scholar and found, for example, 58 studies with the search term “definition of EHR". Five studies were not available to me and by a quick scan 31 studies were invalid or unusable due to the fact that they only stated the search term and not really provided a definition of an EHR. Therefore 22 studies are in my initial set of studies.

Things to do:

·         Step two until four of the used search key

·         Determine other search keys

On March 5^th I have an appointment with Jo Bollen which is a senior general manager working on executive level with more than 25 years of experience in healthcare. Hopefully he can help me further with my master thesis.

Electronic Health Record

Hello everyone,

After consulting Pieter, we decided to shift the focus of my master thesis from the selection process of Hospital Information Systems (HIS) to the selection process of Electronic Health Records (Elektronisch Patiënten Dossier (EPD) in Dutch). We believe that the selection of Electronic Health Records (EHR) is very actual these days and that the field of EHR is more manageable than the field of HIS.

Pieter suggested that I should make more use of my Blog to spread my ideas, even when they are weird, strange or completely wrong. I hope all of you will leave comments on those ideas (in Dutch or English) so I can take those comments into account.

One of those ideas is that we think that the number of required features for an EHR are rather limited, but very powerful. All these basic features are probably present in the current EHR systems so therefore the selection between different EHR systems is not that important. If this is the case we could minimize or even completely eliminate the, often time-consuming and costly, selection process and impose a single EPD system that every hospitals should use in The Netherlands. This would have a huge impact on this market.

Literature review

During the Christmas holiday I really started with the search for relevant literature on my topic. Before that time I already made a start with the literature search but I was very busy with finishing my last master courses. Although those courses are not completed yet, the major part is done so I can start focusing on my literature review.

To be honest it has been proven a bit harder than expected... In the beginning I found that it was difficult to find articles that were both free of charge and relevant to my topic. Lucky enough I realized that the TU/e probably would have an account on the websites so that the articles would be free of charge. Due to making use of keywords I could find relevant articles. On this stage probably I did not find all of them but a good start is made and the Dutch have a nice saying: ‘Al doende, leert men’. ;-)

The main parts on the Hospital Information Systems part that I will try to describe in my literature review are: the history of the HIS, which requirements are contained in HIS, why would hospitals want to invest in HIS, and how are HIS chosen on this moment.

On the selection of HIS part I will try to describe which modeling tools are suitable, (and if for example feature modeling is the appropriate tool) what is feature modeling, which requirements are required for the modeling tool, and which program is the best for the HIS selection.

These are the main parts that I will focus on during my literature review as I believe that they are key issues to address. If other important issues will arise during reading the articles I will takes these into account.


EDIT 12-01-2012:
I have added a tool which allows all of you to subscribe to my blog. Simply enter your email address on the right of the screen and press 'Subscribe', now you have to go to your email account, open the email from FeedBurner and click the link to activate the subscription. Whenever I upload a new post you will receive an email. Please subscribe all!

Kick-off

My name is Michel Koeken and I am an Operations Management & Logistics (master of Industrial Engineering) student at Eindhoven University of Technology, The Netherlands. I started this blog to give anyone who has an affection with my master thesis the opportunity to provide input and feedback.

The topic of my master thesis: Provide a structural approach in the selection process for Hospital Information Systems (HIS­) using Feature Modeling. My aim is to provide hospitals a structural approach which shows them the HIS that has the best-fit according their needs. I plan to make this structural approach via pure::variants. This is a program which allows the user to generate feature diagrams which in the end can be used to make an explicit choice between several HIS.

On this moment I am busy with my literature study and my final master courses. My last exams will be in week 4 and 5 of 2012. After these exams I will focus all my attention on my master thesis. My literature study and research proposal has to be finished before the start of my master thesis in February to make sure that my master thesis will contribute significantly to the existing literature. A few challenging aspects are to master the program pure::variants, to find the relevant decision criteria used by hospitals for a HIS and to acquire the strengths and weaknesses of the different HIS.

Of course I cannot do this on my own and therefore my mentor dr. P.M.E. (Pieter) van Gorp will help and guide me along the way.

If you have any comments, want to provide feedback, offer your help with my literature study or master thesis and/or are interested in my master thesis please feel free to leave a comment which will be highly appreciated.