English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Hospital-based case-control studies employing the “testnegative design” have been used recently, with the intention of reducing confounding by healthcare-seeking attitude in the evaluation of vaccine effectiveness (VE) for some diseases with specific characteristics, such as influenza and rotavirus infection.1–3 Araki and colleagues investigated the effectiveness of rotavirus vaccine for children in a case-control study with this design.1 In this highly sophisticated study, the mechanism of reducing confounding by healthcare-seeking attitude may seem complicated, so it would be helpful for general readers for us to briefly summarize the principle as well as the benefits and cautions related to its application. Figure 1 shows the temporal flow of vaccination and course of development of the target disease; it also summarizes the evaluation of VE in observational epidemiological studies. In a target population, people are vaccinated on a voluntary basis, so that their behavior seeking vaccination might be influenced by their healthcare-seeking attitude. Later, people could be infected by the target pathogens. At this stage, opportunities for infection might be associated with preventive health behaviors that could be associated with healthcare-seeking attitudes. Subsequently, people might or might not develop the target disease, and, whether or not they develop the disease, they might or might not display symptoms of the disease. Symptomatic patients would be likely to visit clinics or hospitals. Other persons would visit clinics or hospitals for other reasons. Some patients without the target disease might display the same or similar symptoms to the target disease; for example, respiratory diseases other than influenza and gastroenteritis other than rotavirus infection. It is a key feature in a case-control study with the test-negative design that there should be such patients and they should be recruited unbiasedly regardless of diagnosis of target disease (explained later). In a prospective study, VE for protection against clinical diseases is evaluated by the ratio of probabilities of developing disease between vaccinated and unvaccinated individuals (Svi=V vs Sui=U). VE is usually expressed as (1 − probability ratio) × 100 (%), so that no effectiveness is expressed as VE = 0% (probability ratio = 1). In usual hospital-based studies, Svi and Sui are approximated by hSvi vs hSui, the numbers of persons who visit clinics or hospitals. The tendency to visit a clinic or hospital (hS=S) may differ between vaccinated and unvaccinated people in accordance with their healthcare-seeking attitudes. Usually hS=S is thought to be larger in vaccinated people than in unvaccinated people, so that the probability ratio would be biased towards 1 (ie, VE would be underestimated). In a case-control study, VE is evaluated as the ratio of odds of vaccinations among cases to the odds among controls (ie, those who did not develop the disease; note that VE = 1 − odds ratio). So, the odds ratio is (Svi=Sui)=((V − Svi)=(U − Sui)) in accordance with the probability ratio in a prospective study. Practical definitions of cases and controls vary by study design, especially for controls. In a hospital-based case-control study, cases are randomly selected from patients who visit clinics or hospitals and are diagnosed as being infected with the pathogens and having symptoms, so that the odds of vaccination among cases is hSvi=hSui. As hS=S is thought to be larger in vaccinated people as aforementioned, the odds tend to be larger than the typical odds of Svi=Sui. If controls are randomly selected from hospital patients who neither become infected with the pathogens nor develop the symptoms of the target disease (ie, hNvn and hNun), the tendency to visit a clinic or hospital (hN=N) is thought to vary according to the illness that led to seeking care. So, the direction of confounding by a healthcare-seeking attitude that is determined by hN=N and hS=S cannot be predicted. In a case-control study with the test-negative design, controls are randomly selected from patients having symptoms that are similar to those of the target disease but who are not infected with the target pathogens (ie, hSvn and hSun). Then, the odds of vaccination among controls are estimated as hSvn=hSun. Here, hS=S in vaccinated patients can be assumed to be the same between symptomatic patients with and without infection (hSvi=Svi = hSvn= Svn, set to be Pv) because patients do not know their infection status before visiting clinics or hospitals. The situation is the same in unvaccinated patients (hSui=Sui = hSun=Sun, set to be Pu). There-

Commentary: Test-Negative Design Reduces Confounding by Healthcare-Seeking Attitude in Case-Control Studies