Administrative data are generally collected based on abstraction of data from medical charts, hospital discharge records, and/or physician billing codes.
Would this type data (i.e administrative data) be considered prospectively or retrospectively collected data?
Would any of the data collected from these sources be considered prospective?
In principle, if the phenomenon, event or outcome that you are studying has already occurred at the time of data collection, it is retrospective data. If baseline data has been collected at enrollment in a prospective cohort study, which records all hospital admissions and discharges, baseline data and hospital discharge data can be used in an attempt to "predict new (not a primary study objective) future events", for example ten years after enrollment. In this case, the study is considered as a prospective study.
Data available on medical slips, discharge slips or medical records usually includes administrative, demographic and clinical/Lab/outcome information. One can abstract the relevant clinical information under study. I agree with the comment of Miikka Korja regarding prospective vis-a-vis retrospective study.
If your event of interest has already occurred and you want to find associated factors with it, you have to traced these back in time-- a retrospective approach.
If you have some baseline information available with you from such records and you want to study occurrence of a new event in future, you have to follow these patients in future for the occurrence of event of your interest-- A prospective approach.
Study of hospital admission rates with event of your interest in series of consecutive years through data collected from hospital, can be considered the prospective occurrence of problem in the given period of time. Therefore it would depend on the investigator, how one want to proceed for what? Though there will be some attached limitations with these types of study. Lack of representation of community, inter-physician variation for diagnosis, lab result variation, lack of a standardized protocol for study etc to name a few limitations.
Thank you all for your input. The reason I asked this question was because I had a discussion with a colleague regarding this issue.
I agree with all your comments as stated above. I think administrative data collected from these sources are retrospectively collected data and suffer from the limitations one would expect when not actively seeking the items in the collection form at the time of interacting with patients. For example, certain variables may be more likely to be recorded by clinicians in the medical record if they carry a particular value.
It entirely depends on when you start your proposal development, project implementation, and timing of data collection!
I think that the retrospective/prospective division is an ambiguous dichotomy.
The important matter is the possibility of controlling bias.
In general we may encounter 3 types of situations:
1- General databases devised to record different kind of data mainly for administrative purposes and being working for a long time.
2- Specific databases, designed with the intention to track information with known importance in order to answer a corpus of questions of quite permanent value.
3- Databases build for the current study and to be filled during the course of it.
Of course, the first option usually implies much more possibilities of bias and confounding factors introduction.
In the last one, the fact that you are following the recording of data makes it possible to detect anomalies when they are occurring without the need to rely on people's memory.
Nevertheless, many other factors could influence data quality. For instance, if the recording process is performed under written rules (like an operating procedure) and/or quality enforcement staff control, the database could have a high data quality.
A related work (not exactly fitting the question but near) comparison of results obtained using randomized controlled trial or case control design could be read at:
http://www.nejm.org/doi/full/10.1056/NEJM200006223422507
Finally, in my humble opinion, anyone involved in knowledge production must use the best data source available at the site with current resources, with a great work at the data understanding and cleansing step, and a good description of this process in the paper. This will allow readers to choose.
Nested case control study data can be collected by retrospective- prospectively, but prospective-restrospective data collection I think not existed in research till date.
- Badges
- Science topic
- Similar topics
- Medicine
- Epidemiology