The Difficulty with Data
This should be simple but its not. For over a year now, I have been tracking my clinic's urine culture collection performance. A while ago, I hit on a methodology that works well for data collection. (In truth, it only works well for me - the data collector). Our prenatal records are scanned onto the computer's public drive after each visit. The records are filed by the Estimated Due Date. For example, everyone who is expected to deliver in September is filed under that month. Data collection for this project goes like this: I go to the public drive and locate the patients who are currently about 20 weeks pregnant (I chose 20 weeks to make sure that everyone was well beyond the 1st trimester). I locate these patients by opening the appropriate "Due Date" month. I have a big Excel spreadsheet where I keep all the data. I record the patient's name, medical record number, month Due and service. If the record indicates that a UA C&S was done, I put a "yes" in the "documentation" and the "done" column. If the record is blank, I put a "no" in the documentation column and go to the EMR to see if a UA C&S was done. In this way, I track documentation and culture rates. Initially, I just used a bar graph to show the results. Finally, I reached a point where I have enough data to use a p chart. Here is where it gets a little confusing. I am looking a patients who are 20 weeks along, but I ID them by the month that they are due and I carry out the analysis using that month, not the month that I recorded the data. For practical purposes, this means that my x axis always includes future dates.
Here's the problem. This data collection and analysis makes perfect sense to me. But, other people don't seem to get it. I present these data to my office staff each month and it took months for me to get beyond the question - "wait, how come the months on the x axis are in the future?" If I need to explain this to my staff month after month, there is clearly a problem. Additionally, I don't know what to do about putting these data in a paper...
So, I am wrestling with a way to re-engineer my data analysis. But, I wonder about doing this. By writing up this project, I am trying to truly represent what I did during the project. How do I acknowledge the limitations of the ongoing data feedback without throwing it out altogether? I don't want to misrepresent what we have done but I want to make the results clear to a casual reader. I am about to do this month's data collection so this is a good time to think about this challenge. One big lesson I have learned. Pay attention when the folks on the front line tell you that they don't get what your data is trying to tell them. It is critically important to make things simple. Otherwise everyone gets lost in the confusion and misses the larger message.