Incorporating group-level exposure information in case-control studies with missing data on dichotomous exposures

Designs for the combination of group- and individual-level data

BACKGROUND

Studies of ecologic or aggregate data suffer from a broad range of biases when scientific interest lies with individual-level associations. To overcome these biases, epidemiologists can choose from a range of designs that combine these group-level data with individual-level data. The individual-level data provide information to identify, evaluate, and control bias, whereas the group-level data are often readily accessible and provide gains in efficiency and power. Within this context, the literature on developing models, particularly multilevel models, is well-established, but little work has been published to help researchers choose among competing designs and plan additional data collection.

METHODS

We review recently proposed "combined" group- and individual-level designs and methods that collect and analyze data at 2 levels of aggregation. These include aggregate data designs, hierarchical related regression, two-phase designs, and hybrid designs for ecologic inference.

RESULTS

The various methods differ in (i) the data elements available at the group and individual levels and (ii) the statistical techniques used to combine the 2 data sources. Implementing these techniques requires care, and it may often be simpler to ignore the group-level data once the individual-level data are collected. A simulation study, based on birth-weight data from North Carolina, is used to illustrate the benefit of incorporating group-level information.

CONCLUSIONS

Our focus is on settings where there are individual-level data to supplement readily accessible group-level data. In this context, no single design is ideal. Choosing which design to adopt depends primarily on the model of interest and the nature of the available group-level data.

The Combination of Ecological and Case-Control Data

Ecological studies, in which data are available at the level of the group, rather than at the level of the individual, are susceptible to a range of biases due to their inability to characterize within-group variability in exposures and confounders. In order to overcome these biases, we propose a hybrid design in which ecological data are supplemented with a sample of individual-level case-control data. We develop the likelihood for this design and illustrate its benefits via simulation, both in bias reduction when compared to an ecological study, and in efficiency gains relative to a conventional case-control study. An interesting special case of the proposed design is the situation where ecological data are supplemented with case-only data. The design is illustrated using a dataset of county-specific lung cancer mortality rates in the state of Ohio from 1988.

Efficiency of two-phase methods with focus on a planned population-based case-control study on air pollution and stroke

BACKGROUND

We plan to conduct a case-control study to investigate whether exposure to nitrogen dioxide (NO2) increases the risk of stroke. In case-control studies, selective participation can lead to bias and loss of efficiency. A two-phase design can reduce bias and improve efficiency by combining information on the non-participating subjects with information from the participating subjects. In our planned study, we will have access to individual disease status and data on NO2 exposure on group (area) level for a large population sample of Scania, southern Sweden. A smaller sub-sample will be selected to the second phase for individual-level assessment on exposure and covariables. In this paper, we simulate a case-control study based on our planned study. We develop a two-phase method for this study and compare the performance of our method with the performance of other two-phase methods.

METHODS

A two-phase case-control study was simulated with a varying number of first- and second-phase subjects. Estimation methods: Method 1: Effect estimation with second-phase data only. Method 2: Effect estimation by adjusting the first-phase estimate with the difference between the adjusted and unadjusted second-phase estimate. The first-phase estimate is based on individual disease status and residential address for all study subjects that are linked to register data on NO2-exposure for each geographical area. Method 3: Effect estimation by using the expectation-maximization (EM) algorithm without taking area-level register data on exposure into account. Method 4: Effect estimation by using the EM algorithm and incorporating group-level register data on NO2-exposure.

RESULTS

The simulated scenarios were such that, unbiased or marginally biased (< 7%) odds ratio (OR) estimates were obtained with all methods. The efficiencies of method 4, are generally higher than those of methods 1 and 2. The standard errors in method 4 decreased further when the case/control ratio is above one in the second phase. For all methods, the standard errors do not become substantially reduced when the number of first-phase controls is increased.

CONCLUSION

In the setting described here, method 4 had the best performance in order to improve efficiency, while adjusting for varying participation rates across areas.