User login
Back to top
Description of the image

UKCCS methods

The UKCCS’s methods and design have been widely reported (1–82). Briefly, the UKCCS sampling frame comprised all children under 15 years of age who were registered with an NHS primary care general practitioner (GP) in England and Wales (1 April 1992 to 31 December 1996) or Scotland (1 January 1991 to 31 December 1994). Cases were children newly diagnosed with cancer and controls were children who did not have cancer. Children were eligible if they were born and resident in Britain in the three months before diagnosis (cases) or pseudo-diagnosis (controls; date that coincided with the exact age at which their corresponding case was diagnosed); a small proportion were excluded because they were adopted or resided in residential local authority care at diagnosis (<1% of the total childhood population).

To identify children newly diagnosed with cancer (cases), proactive notification systems were set-up in all hospitals and specialist treatment centres; and with the permission of the child’s clinical team, their parents were contacted and interviewed. For each case, up to 10 potential control children (matched on sex, year and month of birth) were then randomly selected from the same Family Health Service Authority/Health Board list (in England and Wales or Scotland respectively) on which their corresponding case appeared. Following permission from their GP, the parents of the first two controls (“first-choice” controls) were contacted and interviewed. If the GP refused, or the parents did not wish to participate, another child was selected and the procedures repeated until two control families agreed to take part. In total, the study registered 4430 cases and 9758 controls; 7663 (78.5%) of whom were “first-choice” controls and 2095 (21.5%) were replacements.

Parents (mothers and fathers) who agreed were interviewed face-to face using a structured questionnaire that collected information on a wide-range of potential risk factors. At the end of the interview their consent was sought to abstract information from medical notes (mother’s obstetric and child’s neonatal, as well as mother’s, father’s and child’s primary care); measure levels of radiation (ionising and non-ionising) in their homes; and for parents of cases, to the provision of a blood sample from themselves and their children (cases and siblings). Information about the case’s cancer and treatment were sourced from pathological review of diagnostic material, cytogenetic databases, clinical trials, and hospital records. For all registered cases and controls, the study obtained the children’s birth certificates, and area-based information from the 1991 census using the postcodes of the children’s residence at diagnosis (cases) or pseudo-diagnosis (controls; date that coincided with the age at which their corresponding case was diagnosed).

Linkage to health information contained within national administrative databases is used to track notifications of emigration, cancer and death through the NHS Central Register (NHSCR) in England and Wales, and the General Register Office (GRO) in Scotland. In addition, in order to monitor the longer-term survivorship, the UKCCS links to secondary healthcare datasets (Figure 4) .

Figure 4: Data availability: UKCCS and routinely compiled national sources
Figure 4: Data availability: UKCCS and routinely compiled national sources

Data management and analysis

UKCCS data are managed by the Epidemiology & Cancer Statistics Group (ECSG) at the University of York in accordance with data protection and confidentiality requirements. Study data are stored electronically on a secure computer system that can only be accessed by ECSG staff. To identify individuals in the UKCCS database, anonymised study-specific identifiers are used; no personal information is stored, except the cases’ and controls’ pre-1996 NHS numbers, which are retained only to link to administrative data. The study no longer holds any hard copies of UKCCS data; all paper materials have been destroyed. Data are never stored on laptop computers or portable devices, and secondary healthcare data cannot be accessed remotely via desktop servers.

With respect to data analysis, standard methods are applied; statistical techniques varying with the research question/design. In general, risk estimates are calculated using unconditional logistic regression when studying potential causes using a case-control design, and time-to-event models (e.g. Cox proportional hazards regression) for survival and follow-up events. Analyses that pool findings across studies use individual-level data to estimate study-specific risks before combining these estimates via meta-analysis techniques to give a summary risk estimate. When publishing material, no individuals will be identifiable in the material and all outputs are subject to disclosure control, following the guidelines of the Office for National Statistics.


The Oversight Committee directs and monitors all aspects of UKCCS research. In addition to UKCCS researchers, the committee has an independent chair, patient and public representation, and individuals who bring methodological and/or clinical expertise and rigour. Contributing to the study aims, objectives and interpretation of findings, the committee meets twice a year (or more frequently if required); members are also involved in the development of literature and dissemination of study findings at conferences, meetings and open days, and via websites and social media.