Lesson 6: ART Nested case-control studies

An introduction to nested case-control studies

The nested case-control study is carried out within a cohort study. This lets it share some of the main advantages of the cohort study design.

In this step, we ask you to read a short introduction to the strengths and weaknesses of nested case-control studies and two summaries of eye disease studies which used this methodology. At the end of each summary, we ask you to reflect on some key questions about the study.

Strengths and weaknesses of nested case-control studies

A nested case-control study keeps the advantage of a cohort study in that some exposure data are collected at the beginning of the study when the participant is disease-free. This reduces recall bias, one of the disadvantages of case-control studies. It also keeps the advantage of the
a case-control study in that it is quick and relatively easy to conduct.
Another big advantage of a nested case-control study is that we do not need to carry out detailed examinations on all the disease-free participants, only those we have selected to be controls.

For example, a cohort study of 5,000 participants is followed up for five years and 150 cases of developing disease. The controls are selected from members of the cohort, those did not develop the disease. Due to the size of the cohort, it is very easy to select up to five controls per case.

Study 1: Vitamin D and intraocular pressure – results from a case-control and an intervention study. Krefting EA et al. Acta Ophthalmol 2014

The aim of this nested case-control study was to investigate the associations between levels of serum 25-hydroxy-vitamin D (25(OH)D) and intraocular pressure (IOP).

Method: The study was nested within a population-based study in Norway (Tromso study). All 12,984 members of the population-based study had baseline serum 25(OH)D measured. For the nested study, participants were all self-reported non-smokers.

  • Cases (n=87) had low levels of serum 25(OH)D (between 5th and 10th percentiles)
  • Controls (n=42) had high levels (between 80th and 95th percentiles).

Due to the time difference between the baseline measurements of the cohort and the current study, the researchers took more blood samples and re-analysed serum levels to confirm allocation to case or control status. They also measured IOP of the right eye using a rebound tonometer to exclude those with eye disease or medication that could affect IOP.

Results: IOP in the cases did not differ from that in the controls (15.9 ± 3.3 mmHg versus 15.6 ± 3.1 mmHg, p=0.56, independent t-test).

Conclusion: There was no association between serum 25(OH)D levels and IOP in healthy participants.

Table 1. Baseline characteristics, persons with low and high serum 25(OH)D levels.

Questions for reflection

  • Think about the choice of the study outcome, i.e.what constituted the cases?
  • What biases might have led to the results seen?

Study 2: Exposure to Chlamydia pneumonia infection and age-related macular degeneration: the Blue Mountains Eye Study. Robman L et al. IOVS 2007. This study assessed the cross-sectional and longitudinal associations between exposure to Chlamydia pneumoniae infection and age-related macular degeneration (AMD) using a nested case-control study within the Blue Mountains Eye Study (BMES) cohort.

Method: The BMES I study examined 3,654 persons aged from 49 to 97 years between 1992 to 1994. A second study, BMES II, examined 2,335 participants from the original cohort and 1,174 new individuals (total n=3,509) between 1997 and 2000.

The study we are considering was nested within the BMES II survey and included 197 AMD cases and 433 controls who were matched on age (± five years), sex and smoking status. The cases were those with adequate plasma sample from a total of 349 cases. Eye examinations identified early
and late AMD according to specific criteria. Thirteen controls were subsequently identified as cases and were excluded from the study.

Results: The mean level of antibodies to C. pneumoniae was 0.42 OD units for the 197 cases and 433 controls. After adjustment for age, gender, and smoking status, no significant association was evident between antibody titre (upper versus lower tertile) and any AMD (Odds Ratio 1.02; 95%
Confidence Interval 0.66 –1.56) or by type of AMD.

Conclusions: No association was found between seroreactivity to C. pneumoniae and the prevalence of early AMD. The power of the study would have detected an association with an Odds Ratio of 1.7 or greater. There may have been a smaller association but the study was not
powered to assess this.

Table 2. Characteristics of the AMD cases in the nested case-control study.

* The difference in age between the cases in the nested case-control study (those with adequate plasma sample for testing) and those that were not included was statistically significant.

Questions for reflection

  • Can you think of possible reasons for no association being seen?
  • Do you think the cases should have been matched to the controls by more, or less, variables than the three chosen?
  • Was an association with an OR of 1.7 too optimistic? Why do you think the researchers chose this value?