over in an earlier never ending thread at People of the Eye (latest count for comment entries is 444) there has been a lively and ongoing discussion about Cochlear Implant (CI) safety, ethics, investigation etc
It began due to the Advanced Bionics announced recall on Nov 23 of their HiRes 90K due to some hot failures but the thread advanced and meandered and covered lots and lots of stuff
Joseph Pietro Riolo brought forth some good ideas for some important studies re: CIs and has kindly allowed for us to repost it here in the hopes that a focused discussion of these ideas will ensue.
Thank You Joseph.
I totally agree that it would be great to have a systemic, consistent collection of incidents relating to the failures of cochlear implant so that we can have the total picture. No incident should be dismissed or trivialized, no matter how small it is. However, it is easier said than done. Just imagine how much time, effort and money are required for the enormous task of collecting, documenting and analyzing the information. I will mention three ways of getting information to see how much work each requires.
1. Encyclopedia of incidents. This is somewhat like Wikipedia where each incident has its own page. Each page will explain the incident in details that can give any reader a complete picture of the incident. Then, an analyzer is needed to assign the correct taxonomical (or categorical) code or name to each incident. Problems with this method: People with cochlear implant that fails or that does not work need to be found and once found, they may not be willing to share their experience. It will be difficult to maintain their total privacy if they ask for it. Taxonomy needs to be developed. Analyzers need to be trained so that they can assign right code or name to each incident. Software needs to be developed to collect data and create statistics based on it. There has to be a group of people who must maintain the information on a secure server and protect the integrity of the information. Information needs to be maintained not just for few months but for many years. No matter how much information there is in encyclopedia, it will be always one-sided because no information about the success of cochlear implant is collected.
2. Snapshot of incidents. One way to get a snapshot is to invite all people with cochlear implant to questionnaire. Then, the researchers can analyze the information and provide a snapshot of the non-working cochlear implant. Problems with this method: Questionnaire needs to be developed. Very large number of people is needed to make this successful. For example, if there are 100,000 people with cochlear implant, the study will need at least 10,000 people (10%) to fill out the questionnaire. They would expect total privacy and making sure that no person fills out questionnaire more than once is difficult. Then, there is a problem with sample bias. How can it be assured that the sample is spread over the population equally? Protection of confidential information about them is very important and it will be costly to protect the information. Although this approach can give a good idea of the ratio between success and failure, it is still just a snapshot. The snapshot may lose validity or relevance few years later. So, this study has to be done annually which brings me to the next method.
3. Longitudinal study of incidents. This is same as the previous method but it is done every year. The researchers will be able to measure the changes from year to year. Problems with this method: It will be difficult not to lose any people from each cohort every year. People in the study need to remember to inform the researchers when they move so that they can receive questionnaire every year. Whenever a person forgets or does not want to fill out a questionnaire, a link is broken and it may be discarded. This in turn reduces the sample size which is not desirable. Every year, new people are needed for the study. Questionnaire needs to be developed very well so that it will remain consistent over years. Confidential information needs to be protected at all costs. Storage and integrity of information must also be protected. Very large sample size of new people is needed for every year so that in case some people choose not to participate in the study, the decrease in sample size will not have any adverse effect on the statistics.
As I wrote above, it is easier said than done. Someone or a group of people has to do it.
Joseph Pietro Riolo
XXXXX (email X out here cuz sometimes it creates spam to the blog but u can check Joseph’s comments for his email address)
Public domain notice: I put all of my expressions in this post in the public domain.