Olympiad » Disease
Detectives Event » National
Answer Key: Potential Health Hazards Associated with the Use of Cellular
Problem Set 1
1 point for each, up
to a maximum of 2: ecologic, cohort, cross-sectional, case control
- To minimize study
bias and increase generalizability of results.
- Part (a) and part
- Case definition
- A patient older
than age 18, admitted between 1994 and 1998 to one of three
participating hospitals with primary malignant brain tumor.
(Students can go for 3 of 4 possible points)
- Consult Table 1 below.
Table 1: Characteristics* of persons included in the
investigation of cell phone use and brain cancer
||No Brain Cancer
of hand-held cell phone
|Average daily use
|101 to 499 hours
*Categories within each characteristic are mutually exclusive.
†Some characteristics were not measured for all study participants.
- There is no evidence of a relation between exposure/use and risk
of disease. Some students may conclude a protective effect in the
absence of significance testing.
- Because this is a case/control study, we used an odds ratio to
assess the risk between use of cell phones and occurrence of brain
cancers. Calculations showed the odds ratios to be 1.0 or less,
indicating no increased risk of brain cancers associated with use of
This problem required seven calculations of odds ratios, shown on
the table above. For example, to calculate the risk associated with
average daily use, one would first calculate odds of use more than
215 minutes compared to never/rarely:
Odds ratio (associated with average daily use >=15 minutes) =
ad/bc = (35) (625)/(51)(647) = 0.7.
Odds ratio (associated with average daily use < 15 minutes) =
ad/bc = (103) (625)/(117)(647) = 0.9.
Clarifying note to students and teachers: it is important in
stratified tables such as this that the calculation of odds ratio
always use the referent category. In this instance, the referent
category is "never/rarely" (i.e., the baseline of no
exposure to cell phones).
- Case-control study
- Recall bias due to inability to capture historical patterns of use
or memory impairment due to disease or time lapse. Ascertainment
bias, because people who are more likely or less likely to use cell
phones were differentially included in or excluded from the study
(e.g., due to socioeconomic factors). Information bias, if
investigators’ bias toward expected results influenced data
- Environmental: Long-term (induction periods use, changing phone
Behavior: changing patterns in use.
Back to Problem Set 1