Definitions, rules and standards
The GPCR began data collection in 2004, as a population-based cancer registry. The GPCR registered only primary cancers, with all definitions and protocols of data collection were defined according to standards and rules developed by the IARC and IACR, including multiple primary rules. We aimed to register tumors with malignant or in situ behaviors. In other words, the GPCR did not register tumors with benign or uncertain behaviors. Based on IARC and IACR guidelines, certain patient and tumor characteristics were considered as necessary items for data collection. These included patient’s registration number, first name, last name, sex, age and address as well as topography of tumor, morphology and behavior of tumor, method of diagnosis, date of diagnosis and source of data. The GPCR also collected available data on optional items including patients’ national identification number, father’s name, ethnicity, marital status and phone number as well as grade of tumor, modalities of treatment, vital status and date of death. The GPCR data collection form was developed to ensure the collection of the above-mentioned variables.

Sources of data
We determined all potential sources of data including health care centers throughout the province. This information was obtained from the deputy of treatment affairs of the GOUMS and the list was updated annually. All public and private diagnostic and therapeutic centers (hospitals, pathology/laboratory centers, imaging centers and selected specialist physician's offices) as well as primary health centers throughout the Golestan province were considered as potential sources of data.
The Death registry unit was another source of the data; the GPCR regularly received information on cancer-related deaths from this unit housed within the health department of GOUMS. We matched this data against the incident cancer case file using linkage methods to identify unreported cases; these were then registered as death certificate only (DCO) cases. We also collected information on cancer patients referred to healthcare centers outside the registration area. The GPCR regularly received data from sources in neighboring provinces, particularly in Tehran, Khorasan Razavi and Mazandaran to minimize under-ascertainment.

Data collection methods
The next step was to ensure dedicated cancer registry collaborators were available to support collection from each source. We aimed to find interested personnel from healthcare centers and invite them to collaborate with the GPCR as abstractors; we were able to assign abstractors in 65% of GPCR sources. Among these sources, data were collected passively, i.e., notifications of newly-diagnosed cancer cases were sent to the GPCR office on a routine, and continuing basis. In the remaining 35% of sources for which we could not find an abstractor, active data collection methods were required. GPCR staff regularly visited these sources to directly collect information on cancer cases. All abstractors and GPCR staff were trained in dedicated cancer registry workshops. After training GPCR abstractors and staff, the process of data collection commenced using the GPCR data collection forms.

Coding systems, data entry and analysis
Data on patients' place of resident was collected as a necessary item. At the time of data entry, addresses were coded using an official list of divisions and subdivisions of Iran (including the name of all provinces, cities, towns and villages). We obtained this list from the Statistical Center of Iran (SCI). At the time of analysis, all cases with address codes within the registration area of Golestan province were entered into the analysis and subjects with address codes outside this area excluded.
The third edition of the International Classification of Diseases for Oncology (ICD-O-3) was used for coding tumor characteristics, including topography, morphology, behavior and grade. Data were entered into CanReg-4, a free software created and published by the IARC; the system is being upgraded to the newer CanReg-5; the data were analyzed using the latter software. The GPCR routinely reports indices of the quality of data, as well as the number of cases, age-specific rates, crude rates and age-standardized incidence rates (ASR) using the Segi-Doll world population. In the present paper, as with international projects including CI5, only tumors with malignant behavior were included in the analysis to ensure comparability of the results.