Correct control selection is crucial to the internal validity of case-control studies [1, 2]. The function of controls is to provide valid information on the distribution of exposure within the population at risk of becoming a case [1]. When cases are ascertained from hospitals serving a defined geographic area, a probability sample of unaffected individuals from the population of that geographic region can be used to enhance the likelihood that cases and controls come from the same source population. A concern with this approach is that population controls sampled from a convenient population register may not be representative of the true population at risk of being a member of a hospital case series; population controls may also lack comparability in recruitment fraction and recall of information [1–5]. But there are also concerns that hospital controls, especially those with other diseases, may fail to provide an unbiased sample of the population at risk with respect to exposure status [1–5].

When we applied successfully to Australia's National Health and Medical Research Council to perform case-control studies in China on the effects of green tea on the incidence rates of colorectal cancer, breast cancer and adult leukemia, it became a condition of funding that we recruit both hospital and population controls for around one fifth of the case series to determine if there was any difference of practical importance.

A handful of studies have compared different control groups in western countries where non-emergency access to hospital care generally depends on referral from medical practitioners in non-institutional settings [6–14]. When researchers have enrolled both hospital and population controls, differences in effect estimates using the two groups have varied from inconsequential [6, 7] to problematic [8–10], prompting discussions about characteristics of the exposure, selection protocols and clinical dynamics. Some studies have specifically reported differences between hospital and population controls in key exposure distributions [11–14], but elsewhere the two control groups were found to have very similar exposure distributions [15].

It was important to understand whether exposures measured in the study were in agreement when ascertained from population and hospital control groups, especially because our previous case-control studies of green tea and cancer in China had used hospital controls. This study aimed, therefore, to compare key exposure distributions between hospital controls and alternative population controls to determine if there was any difference of practical importance, and to examine if inferences drawn from green tea effect estimates for different cancer case groups using hospital controls would be different from those using population controls in the Chinese hospital setting.

Data were collected on a total of 1080 controls. Interviews were completed for 540 (89.7%) of the 602 eligible population controls who were approached to participate. The only reason for non-participation was refusal to provide a blood sample, which was an essential part of the study. Of 563 eligible hospital controls who were approached to participate, interviews were completed also for 540 (95.9%). For the 540 recruited cancer cases, the response fraction was 98.5%.

This validation study, as a component of three large case-control studies of malignancies, used separate hospital outpatient and population control groups to assess differences in the distributions of demographic characteristics and lifestyle factors between the two control groups. Furthermore, the study evaluated if different conclusions were reached when comparing cases with matched hospital controls or population controls in the effect of green tea intake. The study found that the distributions of study exposures in hospital outpatient controls were similar to those in population controls. In addition, we obtained similar results in assessing the effect of green tea drinking on three cancers together and separated, using hospital outpatient controls and population controls respectively. Therefore, our results suggest that in the context of a case-control study conducted at a Chinese hospital using urban cases, regardless of whether controls were selected from hospital outpatient attendees without malignancy or drawn from population household registries covering populations in the catchment area of the participating hospital, the two control groups appeared to have similar exposure distributions. Our results identified one exception to this general conclusion. It is possible that population controls may tend to report higher levels of being overweight than hospital outpatient controls.

The findings of this investigation were consistent with a smaller pilot study conducted in 1999-2000 by our research team [20], and with other studies which suggested that population and hospital controls were generally similar in their studies [8–10], as well as with a research by other investigators, who compared non-cancer outpatients with population controls in a case-control study in Japan [15].

A key exposure in our work was green tea drinking measured in three parallel case-control studies. For three cancers together, breast cancer, colorectal cancer, and leukemia, the inversed associations were similar in their direction and size regardless of which control series was employed to represent the underlying exposure to higher green tea consumption. The results, using either type of control series, were consistent with our previous finding that increasing the quantity of green tea consumption reduces the risk of breast cancer in Chinese women [19], as well as the results from four studies pooled in a meta-analysis, which reported a reduced risk of breast cancer for highest versus non/lowest intake of green tea [28]. The association between colorectal cancer and green tea consumption assessed in this study were consistent with another meta-analysis combining results from eight studies for high green tea intake [29]. Although we recruited less leukemia cases, an inverse association was also observed, although it fell short of statistical significance, which was similar to our previous study [30]. The effects of green tea consumption estimated in this study suggest that hospital outpatient controls performed little different from population controls in green tea exposures.

Whilst hospital outpatient and population controls selected in this study may function in a similar manner for a hospital case-control design in China, this does not entirely resolve the underlying theoretical debate as to which of the two types of control group is the more defensible. For BMI, for example, where our study did find evidence of some practical and statistical differences, one is left with the question of which control exposure distribution is more representative of that in the population at risk of becoming a case. The result may have simply been due to chance, although the possibility that population controls were more likely to be overweight cannot be ignored. On the other hand, the lower response fraction of population controls (89.7% vs 95.9% for hospital controls) may have provided the circumstances where normal weight individuals in the general population were less inclined to participate in a research study on health and lifestyle.

In China, public hospitals provide treatment services for acute, severe, and critical outpatients and inpatients, and they fulfill education, research, prevention, and rehabilitation objectives by providing special medical services. All public hospitals have preventive health care branches, which are responsible for reporting infectious diseases, health checks, health counseling, community prevention services, health education, disease screening, family planning and birthing guidance, and care of staff [31]. Patients living in cities readily visit hospitals as non-referred outpatients for check-ups. Cultural factors and insurance arrangements lead patients to maintain a strong relationship with one particular hospital, where they receive a complete range of health care [32, 33]. Survey data in 2008 showed low levels (14%) of community health service utilization, suggesting that community health services are not yet the first point of contact with the health system in China [34]. Therefore, hospital outpatients in China are somewhat similar to ambulatory patients visiting GP clinics in Western countries.

As there is almost never one ideal control group, some researchers believe that population registers provide the most valid way of sampling controls in a hospital case-control study [5, 11–14], because the main theoretical strength of population controls is the potential to provide information of exposure that is unaltered by associations with illness [1]. Others suggest that generally, hospital controls should be preferred in a hospital case-control study regarding practicability, cost and travel time for face to face interviews [9, 10, 15]. There may also be differences in the capacity to recall and report exposures between randomly selected population members and those who are actively engaged in the health system [1]. In addition, only hospital controls have shown some evidence that in the event of developing the cancer, they would be likely to attend the hospital and become a case in the study [1–3].

With respect to the demographic characteristics and lifestyle factors included in the present study, our results suggest that hospital outpatient controls performed little different from population controls for most exposures. For the key exposure, the effect estimates for different cancer case groups using hospital outpatient controls were similar to those using population controls. Therefore, even though some points of concern exist, hospital outpatients provide a satisfactory control group in hospital-based case-control study in the Chinese hospital setting.