Cross-sectional studies

Question 1

Cross-sectional study - design

Answer 1

Sample (or census) of subjects is obtained from the source population, and the presence/absence of outcome and status with regard to putative risk factors (exposures) is ascertained at that point in time.

Question 2

Cross-sectional study - sampling

Answer 2

Ideally, representative sample is drawn through a random sampling process (SRS, stratified, cluster, multistage).

Question 3

Cross-sectional study - types of bias (3)

Answer 3

1. Selection bias: a) non-response bias - if questionnaire is used and responders differ to non-responders; b) survival bias - diseased animal must live long enough to be sampled (underrepresent cases that don’t survive)
2. Information bias: a) recall bias - if exposures are self-reported
3. Confounding bias

Question 4

Cross-sectional study - analysis

Answer 4

Prevalence of disease in the exposed vs unexposed (prevalence risk ratio). Odds ratios often used as logistic regression is frequently used to model multivariable data

Question 5

Cross-sectional study - advantages (4), disadvantages (3)

Answer 5

Advantages

1. Useful for describing baseline characteristics including prevalence (hypothesis generation)
2. Multiple outcomes and exposures can be studied
3. Relatively quick
4. Relatively cheap

Disadvantages:

1. Because cross-sectional studies measure prevalence, and prevalence is related to both incidence and duration of disease, it is impossible to distinguish between those exposures which influnce disease occurance vs disease duration
2. Difficult to investigate cause-effect relationships because exposure and outcome are measured at the same time (best for permanent exposures like breed or gender)
3. Susceptible to bias due to low response/recall issues

Question 6

Cross-sectional study - example

Answer 6

KAP of zoonoses in clients of single vet practice - aims to assess knowledge, beliefs and experiences with zoonoses in relation to type of animal owned, gender, immune status etc

1. External validity: Non-random sample (not representative of all proactices) - findings can’t be extrapolated to other practices
2. Internal validity:
   
   • Selection bias: a) non-response bias could arise if clients chosing to participate differ with regard to exposures vs those who don’t; b) survival bias could arise if certain client groups (e.g. immunocompromised) are underrepresented because they did not survive infection
   • Information bias: a) Recall bias - clients who experience a zoonoses might be more likely to recall a particular incident leading to transmission (bite, scratch)
   • Confounding bias: age may be a confounder in the association betwen education level and zoonosis knowledge