Recall Bias and Quality of Research

Quality research design is essential in public health, health promotion efforts, and epidemiological studies. Lack thereof threatens the internal and external validity of a study. The internal validity of a study is the assurance of equitable representation of subjects within the cohorts groups, the control, and experimental groups (Gordis, 2009). It assures that distribution of the variables or determinant factors for the research is equally distributed within the study. On the other hand, the external validity assures that the sample size selected for a study represents the genetic variability and the socio-economic diversity, etc. of the target population through appropriate selection and randomization of research subjects within the study groups (Gordis, 2009). Perhaps, the appropriateness of epidemiological studies is not limited to a quality research design. The process is a continuous model that accounts for multifactorial variables that can falsely deviate the accuracy of the research questions toward or away from the null hypothesis. Recall bias is one of the multifactorial variables and based on the magnitude of the recall bias within a study; the inconsistency could lead to inaccurate research outcomes of findings.

A recall bias affects the internal validity of a case-control or retrospective cohort or prospective cohort research design (Hassan, 2006). Recall bias is the tendency of reporting past events inaccurately, thus resulting in a differential misclassification of human subjects within or between study groups (Hassan, 2006). Essentially, differential misclassification of subjects is the tendency of assigning human subjects to the wrong group in a study due to information bias obtained either from subjects or other means (Gordis, 2009). Recall bias is an example of information bias.

Recall of information from human subjects employs memory trace to retrieve information from the brain. Scientific studies showed that human memory traces are the poor version of original percept (Hassan, 2006). Scientific evidence also showed that within 1 year of a critical event about 20% of the information are irretrievable from the human brain, and about 50 % are irretrievable after 5 years (Hassan, 2006). Hence, recall bias is a crucial factor to consider and assess while conducting a case-control, prospective, and especially, a retrospective study.

Based on Sorensen (1989) assessments on the effects of alcohol on liver injury (Cirrhosis) in a case-control, prospective and retrospective studies, it is evident that information bias and confounding factors play a role in deviating the research outcomes or findings. The clinical consensus is that Cirrhosis develops within 12 months of abstinence from alcohol abuse while severe liver injury, at a minimum, occurs within 3 months preceding the time of abuse (Sorensen, 1989). The timeline was a crucial factor necessary in investigating the alcohol effects on cirrhosis and liver injury. It is also a critical factor in validating subjects’ alcohol abuse 1 year prior, and ascertaining the quantity consumed in the last 3 month before abstinence.

The case-control assessment described by Sorensen (1989) indicated that Cirrhosis etiology is associated with alcohol abuse. The associations were dependent on the dose of alcohol consumption. The “alcohol dosage” to “effects” described in the study ranges from 20g to >160g and from 20g to >80g of alcohol consumption (Sorensen, 1989). Based on the inconsistencies of the quantitative range of alcohol consumption that leads to Cirrhosis, the author admitted that there are possible confounding factors associated with the onset of Cirrhosis among “alcohol abuse” subjects.

In a cohort prospective assessment, 15.5 % of subjects with Cirrhosis were confirmed via biopsy while 14.2% of subjects with Cirrhosis were confirmed with other clinical evidence/tests (Sorensen, 1989). The study showed that the case of developing Cirrhosis within the studied group was statistically independent of the duration of alcohol abuse (Sorensen, 1989). The conclusion was based on the fact that Cirrhosis onset extends from 1 year to 30 years. In addition, the reported daily intake was reported as 50g to >250g (Sorensen, 1989). Based on the information of dosage effect, the study suggested that there is no long-term or cumulative effect of alcohol-related Cirrhosis. The study showed that Cirrhosis attains a stable level at a daily intake of 75g of alcohol, which questions the role of “risk of functions” such as genetic or social predisposition, etc. to the expected health outcome (Sorensen, 1989).

On the other hand, the cohort retrospective assessment described in Sorensen (1989) article indicated a 0% to 51% range of Cirrhosis cases among subjects who participated in the study. In the study, the dosage intake associated to Cirrhosis was >160g of alcohol consumption. However, when two groups with similar dosage below 160g of alcohol were used to correct the prevalence of Cirrhosis in the group, the prevalent was adjusted to 14% instead of 0% (Sorensen, 1989). The variation is indicative of recall bias or that subjects over reported the level of alcohol consumption due to rumination bias and other factors.


Gordis, L. (2009). Epidemiology 4th ed. Philadelphia: Saunders Elsevier.

Hassan, E. (2006). Recall bias can be a threat to retrospective and prospective research designs. Internet Journal of Epidemiology, 2(3), 4-4. Retrieved from %40session mgr111&vid=1&hid=127&bdata=JnNjb3BlPXNpdGU%3d#db=a9h&AN= 22343242

Sorensen, T. (1989). Alcohol and liver injury: Dose-related or permissive effect. British Journal of Addiction, 84 (6), 581-589. Retrieved from /ehost/detail?sid=4e41b532-697f-4cce-9c8f-771bd484b0c8%40sessio nmgr114&vid= 1&hid=127&bdata=JnNjb3BlPXNpdGU%3d#db=a9h&AN=6613453