I have to back up what Nell says here. I am all fine and dandy with supporting research, but just knowing A1c and perhaps the minimal profile information disclosed here is not sufficient to conduct any real research. I don’t have any problem showing support for efforts, but I medical studies find it hard enough to sift through mounds of confounding variables and not being able to measure or control any variables really makes the value questionable. Most medical research makes special effort to properly and legally protect patient information, there has been no mention of legal protections or HIPAA and I am not sure whether it applies and it does not seem that anyone is being informed properly. I have not revealed much of my private information here for the public and while I don’t have any problem doing that for research, I want protections.
Just looking at my picture you can imagine how much I worry.
There are several reasons why I do not share. One is that I am concerned about privacy. The other is that I know nothing of the program which cropped up when I was rather ill and, I think, in the hospital this last time. I do not access the internet often enough lately because it is hard for me and they home is not sharing their WIFI! When I am sick, I don’t fiddle around with strange things.
I added mine on one of the first days that it went up. I did it because I thought it was a fun thing to do. I know I am bad, but I am also a little bit competitive so thought it would be fun to see how my state did compared to others. I am disappointed that more people from PA haven’t participated since we were close a week ago and still not there.
Thanks a lot to Nell and bsc for sharing your thoughts about TuAnalyze so candidly.
I want to address a portion of the comments made in connection with privacy. I have also asked the Children’s Hospital Boston team to post in here about the research questions posted, so that you can also hear from them (not just me) in regards to that.
In the “About” tab of the TuAnalyze map, you can read (at the bottom) the following:
“TuAnalyze conforms to the TuDiabetes privacy policy and to the ethical and regulatory requirements of the Children’s Hospital Boston Institutional Review Board.”
The data gets displayed/shared on TuDiabetes in direct accordance to the sharing settings you choose. So if you choose to share “all the way” (meaning, letting anyone who can see your profile, based on your profile privacy settings- see your data) that is what is done. If you choose to not share at all, your data is de-identified and anonymized, looking something like “A member of TuDiabetes (someone among the 15,000 members) entered an A1c value of X.”
To give you an idea of how seriously they take this, even I was required by their Institutional Review Board to complete an online training on Research Ethics (which I confess was enlightening but not the most entertaining subject I have studied -he, he!) Just for referential purposes, I am attaching my completion report of the training below.
Anyway… this addresses some of the questions brought up. Please do not hesitate to let me know if you any of these answers still doesn’t address privacy-related concerns to your satisfaction.
I will defer to the Harvard team to answer the research questions. 6426-CompletionReport.pdf (94.2 KB)
Lois,
Besides the reply I just posted at about the same time you posted, I thought you’d want to check out this video:
(hope the still of me with my eyes closed and mouth open doesn’t scare you or anyone else out!)
As one of the principal investigators of the TuAnalyze project, I am following this thread with interest. By way of introduction—my name is Elissa Weitzman. I am an Assistant Professor at Harvard Medical School and Children’s Hospital Boston.
The questions about research raised in this thread are astute. The first goal of this research, using TuAnalyze, is to create a working model for collecting information from and for the community that can be used for public health research. The model should be open, provide for consent, and flexible. We chose to undertake a first test around A1c because it’s an indicator that many people are familiar with, has fairly clear meaning in terms of diabetes, and was already being discussed in the community. As we learn from this experience, we look forward to collecting other data–hopefully without burdening the community–with which to analyze the A1c data and other measures.
Everyone here is correct that traditional models of data collection for population monitoring of health are highly structured and rely on collection of extensive information–through clinic visits, mailed surveys, or interviews. A great deal has been accomplished using these methods. We won’t replace them, but we are trying to augment them with a different model. So the path is:
Step 1: launch, listen, and learn about the approach
Step 2: compare what’s learned here with what’s learned using traditional models on a first indicator (A1c) and from there, move on to other issues and measures
Step 3: innovate, and add to what is known, recognizing that the flexibility of this technology once connected to a large number of people may help move the science of diabetes research forward.
As we move along this path, we aim to provide opportunities for deeper engagement in research (voluntary) and make a broader set of tools and features available to the community.
It’s a privilege to do this work! Thanks for participating and asking good questions.
from Elissa Weltzman (excerpt)
Everyone here is correct that traditional models of data collection for population monitoring of health are highly structured and rely on collection of extensive information–through clinic visits, mailed surveys, or interviews. A great deal has been accomplished using these methods. We won’t replace them, but we are trying to augment them with a different model. So the path is:
Step 1: launch, listen, and learn about the approach
Step 2: compare what’s learned here with what’s learned using traditional models on a first indicator (A1c) and from there, move on to other issues and measures
Step 3: innovate, and add to what is known, recognizing that the flexibility of this technology once connected to a large number of people may help move the science of diabetes research forward.
[end excerpt]
My response:
Sorry, but this is not a response that helps those of us who asked about “what research?” It appears that, reading deep between the lines, you are collecting A1c data as a test to see if you can collect data from a community like this. Period. You will do nothing with the A1c because it is an easy data point to collect to see what is the response of the online community. (right, so far?) Then, if you get a large enough sample from the population of interest, you will plan a study and come back for more complete data, no doubt including A1c, demographics, and other variables. If this is the case, or if something else is the case, this all should have been clearly explained before starting the tuanalyze data collection. It is not appropriate to ask people for one data point and just say that it is “for research.” Since you indicate you are following institutional and federal IRB guidelines, you will know that a clear explanation of what the data will be used for is a requirement.
I am not trying to be hard-nosed. I am all for research. But I do think it is not ethical to mislead people into thinking that some outcome of interest will result from their sending in their A1c numbers. Unless I read your response wrong, it is just a test as I noted previously. Nothing wrong with that as a starting point for later research, as long as it is explained up front. You could as easily asked for age and accomplished the same thing, as I understand it. Of course, the A1c is of more personal interest to diabetics. Again, I am not trying to discourage anyone from sending in their A1c as I see no harm in doing it, especially as described. But I don’t think anyone should expect a research outcome from it.
I agree with you, Nell. This TuAnalyze seems like a fun social networking gimmick and not research. I don’t know how scientific this data could be. My guess is that more people with low A1C’s will post than people with high A1C’s for several reasons, including privacy concerns, and the fact that anyone taking time to join tudiabetes, follow diabetes research or post their results on this study probably manages their diabetes a little more intensively. Plus, the highest A1C’s are the ones never recorded. I have not looked at the map or the results so I may be wrong, but I predict that the results would be skewed toward lower A1C numbers.
Hi Nell,
You raise solid points including what is “research”. I think it would be ambitious of us to try to find an answer to something within the first 2 weeks of launching TuAnalyze although all of us, I think, have a short list of burning questions. We took a specific, limited approach to opening this up. A different approach would have been to design the app for member reporting about an extensive list of data points, with perhaps a greater immediate burden to members, and no guarantee that anyone would use it and no ability to predict the extent of participation bias. This is important to note because the burden on the investigator is to try to balance risks/benefits. If there is a place in the app materials (i.e., the tabs and the preferences and notification) where we can alter the text to read more clearly that this project will develop and mature to address a range of scientific questions, we are all for it and welcome your advice.
Thanks for the clarification. I am pleased to see you did so well on the quiz. The privacy issue I am bringing up is something that many members have not thought about. From the tudiabetes privacy faq:
Members are not required to provide medical/health information to use our Service. If however, you choose to provide medical/health information, please remember that any information that you disclose becomes public information.
This means that information you have posted here is in the public domain. As to whether this is “informed consent” is perhaps an open question, but in either case you have gone to some effort to put in place a specific form asking users to formally consent to sharing.
I’ll just say, that in my opinion, posting of presumably “Private Health Information” in a controlled social networking site does not necessarily satisfy “informed consent” to the arbitrary use of that information.
Please understand. I hide behind Alfred’s face, I don’t give out much personal detail unless I know you. I trust you, but I don’t trust everyone. It is but a short hop to this information being used by insurance companies.
Thanks for providing more information. I guess I still have real questions about what this is all about. Is about some sort of new model for participatory research? That is fine, but as nell has pointed out traditional study methods have focused on much more structured and controlled information collection. This may be very innovative, but as you are aware I am sure, just collecting HbA1cs by geography without much control is unlikely to add much to our knowledge. When you discuss your work, you might think about talking about finding new associations, environmental triggers or things like that. These are things that are unlikely to be found in traditional hypothesis driven information collection, but might be found in social networks.
I will consider contributing my information. Thanks to you and Manny for hearing out my concerns and responding.
The statement in our Privacy Policy that you quote refers to things members may post in the Forum and other public areas of the site since those are open spaces and are therefore subject to public view.
I will draft a statement that clarifies that data entered through portions on the site such as TuAnalyze is subject to the privacy/sharing settings you choose for your submission. Indeed, you also have a similar option when you post a blog entry or upload a photo as well as in connection to your own profile page.
Thanks so much for the feedback. With thoughtful comments like yours we can make sure the information is clear and conveys exactly what is happening in terms of privacy/data sharing.
One idea I am going to throw out there, which indeed can be established through the kind of data collection under way in TuAnalyze is something like this:
Even assuming that mostly people with low A1c numbers are the ones reporting their numbers, the trend stemming from the first A1c number can still be determined and be useful.
Imagine that trend shows an improvement (or not): whatever the outcome of the trend…
Then think of behavior patterns that members in the network may exhibit: how many friends they have on average, or how active they may be.
Now, there could be correlations assigned between such data trends and such social network behavior patterns. With those correlations hypothesis could be created that could lead to further studies to try to verify whether INDEED there is more than a correlation (if there is causation) between the two.
This is just an example of something that has always been an exciting possibility that could open up doors into knowledge that could help millions of people depending on what we learn from it.
I tried to enter the correct date and it would not accept it. It only showed the calendar for June. My next HgA1C is due now so I will enter it as son as I can.
Mine was in Jan & I was able to enter it. When the June calendar pops up, there is an arrow on the left side of the gold bar that has “June.” Just click on that several times until you reach the month you need.
Good point, Greg. Although I have used interview research techniques with good results, in general, many researchers consider self report to be biased. To be clear though, all research, even basic, has biases. The trick is to recognize and deal with them!
Dianne,
Can you please let me know what browser you are using? Ideally if you can use the Contact Us form at the bottom of the page to report this, that will include the browser version along with other technical details about your browser configuration that can help us troubleshoot this.
bsc,
This is how the Privacy Policy has been clarified to make this information clearer:
“Members are not required to provide medical/health information to use our Service. If however, you choose to provide medical/health information in public areas of the site (Forum, public Groups, etc.) this information becomes public information. Medical/health information entered entered in areas of the site that offer privacy/sharing settings (such as the TuAnalyze application, your profile page, your blog, photos and videos) are subject to the sharing settings or privacy settings you choose for your submission.”
Do you think this makes the distinction easy to understand?
