Question

Indicate the one major type of limitation that is possible in each of the situations described below.  Choose from the following: selection bias, information bias, confounding, random error, limited generalizability or no error.  Choose only one per example.

1. A case-control study was conducted to determine whether lower socioeconomic status was associated with a higher risk of cervical cancer.  The cases consisted of 250 women with cervical cancer who had been referred to Massachusetts General Hospital for treatment for cervical cancer.  They were referred from all over the state.  The investigators identified controls by going from door-to-door in the community around Massachusetts General Hospital from 9:00 a.m. to 5:00 p.m.  Many residents were not home, but they persisted and eventually enrolled enough controls.  
2. In a prospective cohort study of risk factors for disability, 25% of the sample has been lost to follow-up throughout the 10 years of the study.  It was discovered by reviewing the baseline (start of the study) data that those lost to follow up were twice as likely to have risk factors for disability as the 75% who remained in the study.
3. A study of psychiatric disorders found a positive association between low socioeconomic status and psychosis (severe mental disorder).  It also found a strong negative association between low socioeconomic status and neurosis (mild mental disorders or emotional problems).  It turns out that those of high socioeconomic status were also more likely to seek treatment for emotional problems, leading to a diagnosis of neuroticism, than were their lower SES counterparts.  Many neurotic people in low SES groups did not have the luxury of accessing treatment, and were simply misclassified as normal.
4. An epidemiologist reported the following conclusion from a large study during a national meeting: “Daily use of 500 milligrams of vitamin C for a period of one year was associated with reduced frequency of upper respiratory infections in children under 10 years of age.” (Relative Risk = 0.56, 95% Confidence Interval = 0.11- 4.21).
5. Results of a study were published announcing that coffee drinking was a risk factor for bladder cancer.  However, it was later noted that coffee drinkers were much more likely to smoke than non-coffee drinkers, and that smoking is a risk factor for bladder cancer.  When multivariate analyses were done, adjusting for smoking, coffee drinking and bladder cancer were no longer associated.
6. In a case-control study of prostate cancer and sexually transmitted diseases, the cases with prostate cancer had a more detailed memory of their exposure to STDs in the past than did the healthy controls.
7. Investigators wanted to study the association between a person’s likelihood of quitting smoking and the smoking status of one’s intimate partner.  A sample was used that consisted only of heterosexual couples who were legally married.
8. To investigate whether body mass index is associated with high blood pressure, 2,165 middle-aged men and women were randomly selected in Oslo, Norway, in 1979.  Blood pressure, height, and weight were measured by trained and certified personnel who used standardized, objective procedures that had been developed and tested previously in a large, multi-center study of treatment for hypertension.
9. Given that the risk of coronary heart disease increases with age, a cohort study of the association between physical activity and coronary heart disease was conducted in which the average age of the physically active group was 10 years younger than the average age of the inactive group.
10. The respondents to a cross-sectional survey may have answered highly sensitive

questions in the ways they thought the interviewer would expect them to answer.

11. A study was conducted of 3,000 military troops deployed at an atomic test site in Nevada to detect the occurrence of leukemia.  A total of 1,870 persons were successfully traced by the investigators, and an additional 443 contacted the investigators on their own as a result of the publicity about the study.  Four cases of leukemia occurred among the 1,870 individuals traced by the investigators, and an additional four occurred among those individuals who contacted the investigators on their own.
12. A survey is done to determine the prevalence of carpal tunnel syndrome in employees.  None of those who were interviewed were eligible for disability compensation.
13. An association between high birth order (3^rd, 4^th, 5^th or later child) and Down’s Syndrome is found, but infants of high birth order are more likely to be born to older mothers, who also are more likely to give birth to infants with Down’s Syndrome.
14. A study was conducted which investigated the possible association between coffee drinking and oral cancer.  No association was found.  The investigators failed to take account of two well known risk factors for oral cancer, smoking and alcohol consumption.  Both smoking and alcohol consumption are also more common among heavy coffee drinkers.  When a later study did control for these two risk factors, coffee drinking was found to have a protective effect against oral cancer.
15. A random sample of 10 people was drawn to represent the City of Chicago.  These people were then questioned about their health behaviors.

Answer 1

A case-control study was conducted to determine whether lower socioeconomic status was associated with a higher risk of cervical cancer.  The cases consisted of 250 women with cervical cancer who had been referred to Massachusetts General Hospital for treatment for cervical cancer.  They were referred from all over the state.  The investigators identified controls by going from door-to-door in the community around Massachusetts General Hospital from 9:00 a.m. to 5:00 p.m.  Many residents were not home, but they persisted and eventually enrolled enough controls. It is the example of selection bias as women from all over the state were referred there but researcher took control only from the community around the hospital.

In a prospective cohort study of risk factors for disability, 25% of the sample has been lost to follow-up throughout the 10 years of the study.  It was discovered by reviewing the baseline (start of the study) data that those lost to follow up were twice as likely to have risk factors for disability as the 75% who remained in the study. It is the example of selection bias due to lost to follow up

A study of psychiatric disorders found a positive association between low socioeconomic status and psychosis (severe mental disorder).  It also found a strong negative association between low socioeconomic status and neurosis (mild mental disorders or emotional problems).  It turns out that those of high socioeconomic status were also more likely to seek treatment for emotional problems, leading to a diagnosis of neuroticism, than were their lower SES counterparts.  Many neurotic people in low SES groups did not have the luxury of accessing treatment, and were simply misclassified as normal. it is the example of confounding bias as here extraneous factor of not seeking treatment by lower SES for neurotic disorder affect the interpretation.

An epidemiologist reported the following conclusion from a large study during a national meeting: “Daily use of 500 milligrams of vitamin C for a period of one year was associated with reduced frequency of upper respiratory infections in children under 10 years of age.” (Relative Risk = 0.56, 95% Confidence Interval = 0.11- 4.21). It is example of limited generalisability as results are limited to children under 10 yrs

Results of a study were published announcing that coffee drinking was a risk factor for bladder cancer.  However, it was later noted that coffee drinkers were much more likely to smoke than non-coffee drinkers, and that smoking is a risk factor for bladder cancer.  When multivariate analyses were done, adjusting for smoking, coffee drinking and bladder cancer were no longer associated. It is example of confounding bias as smoking as confounding factor affects results.

In a case-control study of prostate cancer and sexually transmitted diseases, the cases with prostate cancer had a more detailed memory of their exposure to STDs in the past than did the healthy controls no error.

Investigators wanted to study the association between a person’s likelihood of quitting smoking and the smoking status of one’s intimate partner.  A sample was used that consisted only of heterosexual couples who were legally married. Example of limited generisability to legally married homosexual couples.

To investigate whether body mass index is associated with high blood pressure, 2,165 middle-aged men and women were randomly selected in Oslo, Norway, in 1979.  Blood pressure, height, and weight were measured by trained and certified personnel who used standardized, objective procedures that had been developed and tested previously in a large, multi-center study of treatment for hypertension.no error

Given that the risk of coronary heart disease increases with age, a cohort study of the association between physical activity and coronary heart disease was conducted in which the average age of the physically active group was 10 years younger than the average age of the inactive group. Random error as sample not randomised accurately

The respondents to a cross-sectional survey may have answered highly sensitive questions in the ways they thought the interviewer would expect them to answer. Information bias as the information given by participants is modified.

A study was conducted of 3,000 military troops deployed at an atomic test site in Nevada to detect the occurrence of leukemia.  A total of 1,870 persons were successfully traced by the investigators, and an additional 443 contacted the investigators on their own as a result of the publicity about the study.  Four cases of leukemia occurred among the 1,870 individuals traced by the investigators, and an additional four occurred among those individuals who contacted the investigators on their own. Selection bias

A survey is done to determine the prevalence of carpal tunnel syndrome in employees.  None of those who were interviewed were eligible for disability compensation. no error

An association between high birth order (3rd, 4th, 5th or later child) and Down’s Syndrome is found, but infants of high birth order are more likely to be born to older mothers, who also are more likely to give birth to infants with Down’s Syndrome.confounding bias as extraneous factor of old age of mother affects results.

A study was conducted which investigated the possible association between coffee drinking and oral cancer.  No association was found.  The investigators failed to take account of two well known risk factors for oral cancer, smoking and alcohol consumption.  Both smoking and alcohol consumption are also more common among heavy coffee drinkers.  When a later study did control for these two risk factors, coffee drinking was found to have a protective effect against oral cancer. Information bias as interpretion if result carried out inaccurate

A random sample of 10 people was drawn to represent the City of Chicago.  These people were then questioned about their health behaviors. limited generisability because of limited participants