html xmlns=”http://www.w3.org/1999/xhtml” xmlns:mml=”http://www.w3.org/1998/Math/MathML” xmlns:epub=”http://www.idpf.org/2007/ops”>
Although blinding is typically thought of as something done in interventional studies, it can and should be used in observational studies as well. This reduces the chance of biasing the results, since judgment is needed when extracting or collecting data to be used in observational studies. There are multiple ways that investigators may be blinded to aspects of the participant’s history to reduce bias. We describe these first for written records, then situations in which participant contact is necessary, and finally for the situation where only a single investigator is involved in a study, which is typical in a record review.
28.1 What Can Be Blinded?
The standard for interventional studies is the double-blind study, in which neither the participant nor those involved in assessing the outcome know which intervention the participant is receiving. This is not possible in an observational study, since the participant knows at least the exposure or the outcome, and often both. Thus, the focus of blinding in an observational study is on blinding those involved in the assessment of the participants. In an ideal world of scientific research, all exposures and outcomes would be measured by precise and reproducible criteria and would not require subjective judgment. Unfortunately, we do not live in such an ideal world, and so the protocol needs to specify how, precisely, blinding is to be done to minimize the potential for bias in the results. There are three basic features of a study that could be blinded to those involved in the data abstraction and collection: the exposure information, the outcome information, and the actual hypothesis underlying the study that links the exposure and outcome information.
28.2 Blinding of Written Records
When only written records will be used, different assessors can be used to review the exposure information and the outcome information. In the simplest situation, one set of records would contain the exposure information and another set of records would contain the outcome information.
This is a cohort study of individuals who worked in a chemical plant during 1970–1990. The objective was to determine if workers who had “high” exposure to a specific chemical had a reduced respiratory function at age 60 compared to workers in the same factory who were either not exposed to this chemical or had only short-term or minor exposures. The raters reviewed the work records of each individual and classified them as high, moderate, or low exposure based on their job description, work areas, and time in these jobs. Different raters reviewed the participants’ medical records for signs of deterioration in respiratory function. The raters reviewing the work records were blinded to the participants’ outcomes; the raters reviewing the medical records were blinded to the participants’ exposure.
Often, however, there is not a clean separation of the exposure and the outcome information. In that case the documents must be copied and the information must be redacted, either by blacking out the information with a heavy pen or cutting away that part of the document. Two sets of documents are needed: one with the exposure information redacted and one with the outcome information redacted. There should be a designated person involved in doing this task, and this person should not be involved in the subsequent data abstraction of either the exposure or the outcome information for any of the records they reviewed in full. Like in Example 28A, there should then be different assessors for the redacted exposure information and the redacted outcome information, to minimize the possibility of assessor bias when abstracting the data.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
