The KCR uses a combination of passive and active approaches to data collection. In the passive method, certain health care providers (pathology laboratories) report data on cancer patients in their care to the registry. In the active method, cancer registrars visit health care facilities, such as Mulago Hospital, a 900-bed National Referral hospital with several units, to identify cancer cases. Active data collection is needed because Uganda does not have laws requiring health care facilities to report cancer cases to registries, but the KCR has some agreements in place for providers to allow data abstraction. The registry employs two registrars who travel to five other hospitals in Kyadondo County to obtain data at least once every 2 months. During these visits, the registrars consult the admission and discharge registers, clinical notes, and pathology reports. In addition to the two cancer registrars, record assistants in various units within the hospitals assist in extracting data from the patient records. The registrars also routinely visit the Uganda Hospice to identify cancer patients. Certification of death is only carried out for legal reasons and is very incomplete, so the registry only uses death certificates issued in the Mulago Hospital mortuary as a source of information. The registry performs data management using the CanReg software [11], which includes checks for consistency and validity. The software also allows for searches to identify potential duplicate registrations. Only authorized personnel can access the registry data, and confidentiality is maintained by using only registration identifiers during data analysis.

We used the IntRegCosting Tool to estimate the resource use by budget categories and to estimate the activity-based cost [12]. The IntRegCosting Tool consists of 10 modules that describe types of information, such as budget categories or other factors, and 30 activities that describe specific registry functions. The IntRegCosting Tool modules and activities have been described in detail previously [12]. The registry activities and definitions were tailored in consultation with registry staff, and matched to the categories used by other registries in low- and middle- income countries who pilot tested the IntRegCosting Tool. The registry provided information on labor and non-labor resources for each budget component and then assigned costs to 30 specific registry activities. Cost data were collected for 2014. The registry was able to accurately assign more than 95% of the contributions received to specific registration activities; the KCR was able to provide high-quality activity-based cost data using the criteria previously established for high-quality cost data allocation greater than 90% [13,14].

We determined the number of incidence cancer cases at the KCR from 1991 through 2012 and computed age-standardized incidence rates for the most recent 5-year period (2008–2012) using the direct method with the world standard population [15].

The number of cancer cases recorded by the KCR has continually increased over time, mainly because of growth and ageing of the county population as an increasing number of individuals come from rural to urban areas to seek work and better standards of living, but also because of increasing incidence rates for some of the major cancers [4].

The GCR is in the initial stages of population-based data collection, and therefore, only preliminary results, based on the 656 cases registered in 2013–2014, are available. Age-standardised rates were calculated as well for Kampala.

We analyzed resource use and cost data using Microsoft Excel. We categorized the cost of the registry for each budget category, including labor, travel, equipment, and administrative or indirect costs. We summarized labor and non-labor information from the Microsoft Excel–based IntRegCosting Tool to determine the cost of specific registry activities.

We categorized the 30 registry activities into fixed and variable cost activities. By definition, fixed costs do not vary as the volume of cases changes (at least in the short run), while variable costs do change. The variable cost–related activities were further subdivided into core activities – those that were essential for registry operations – and other activities, such as advanced analysis (e.g., investigating disease clusters) and research-related tasks. We also report the cost per cancer case by activity type, which we calculated by dividing the total value of registry resources towards fixed cost, variable cost core, and variable cost other activities each by the average number of incident cases registered during a typical annual period (we used the 1907 incident cancers diagnosed in 2012).

As stated in the introduction, the GCR was in its start-up phase at the time of the study. To project the possible cost of operating the GCR, we qualitatively compared key characteristics of the KCR and GCR. These characteristics included population served, size of the coverage area, number of data sources, data collection approach, and whether the catchment area had a cancer referral center, all factors previously shown to impact the cost of registry operations [12]. Using the differences in these factors, we identified whether the cost of specific GCR operations was likely to be lower, higher, or the same as the KCR. We also identified a list of activities and potential sources of monetary and non-monetary contributions to sustain the GCR operations in the future.

Table 1 summarizes the key characteristics of the KCR and GCR. The population served by the GCR, about 756,547, is a little less than 30% of the population of the Kampala registration area, but the area covered is about six times larger, about 11,820 square kilometers across three Ugandan districts [16]. With a smaller population over a much larger geographic area, the GCR’s population density is 64 persons per square kilometer, which is a contrast to KCR’s 1374 persons per square kilometer. While Kampala uses a combination of passive and active approaches to collect data, the GCR relies on active data capture to collect information from Lacor Hospital and five other facilities in the Amuru and Nwoya Districts. Kampala has more data sources than Gulu and is also home to a major referral cancer treatment hospital. As shown in Table 1, the GCR serves a smaller population and has fewer data sources; these factors lessen the effort required to collect data.

Since its inception, the KCR has instituted mechanisms for ensuring high quality data. These mechanisms include continuous retraining of its registrars on data collection methods and processing. Furthermore, the registry undertakes completeness assessments of the cancer cases by using a large case series collected independently of the cancer registry mechanisms. The completeness of registration using this approach was assessed as 90%, indicating a generally high level of accuracy in the incidence rates reported by the registry [17]. Another method used to determine quality is matching studies, such as the Uganda AIDS-Cancer Registry Match study [8]. The matching process indicated that the KCR was able to provide good quality variables for the linkage process and for the identification of specific cancer sites. KCR data have been included in Cancer Incidence in Five Continents (CI5) continuously since the 1990s due to the registry’s rigorous quality assessment approach to ensure high quality data. Results from the KCR have been published in volumes VII (1991–1993), VIII (1993–1997), IX (1998–2002), and X (2003–2007) of CI5.

Table 2 shows the top five age-standardized incidence rates for Gulu in 2013–2014 compared with those in Kampala in 2008–2012. Overall, Gulu age-standardized incidence rates are lower than the rates for Kampala, although the incidence of cancer of the cervix is higher. In part, this difference in rates reflects the more rural nature of the population in Gulu. The incidence of childhood Burkitt lymphoma is also higher in Gulu than in Kampala [9].

The main sources of support for the KCR were the host institution (Makerere University), the International Network for Cancer Treatment and Research (INCTR), and the International Agency for Research on Cancer (IARC). The host institution, which provided 58% of KCR’s total resources, contributed through direct transfers (e.g., salary for one of the registrars) and through other contributions, such as use of facilities, provision of services (e.g., administration, IT support, and transportation), equipment (e.g., computers and printers), and furniture. International organizations, which included INCTR and IARC, contributed 42% of KCR’s resources.

Fig. 2 presents the cost distribution of the KCR by the standard budget categories of labor, travel, training, equipment and materials, and administrative costs. More than half of the costs were related to registry staff salaries and other labor compensation (59%), and 12% were for administrative support related to registry space, information technology, and utilities. Equipment and materials, such as office supplies, hardware, and printing, accounted for 15% of the registry’s costs. Travel and training-oriented non-labor payments were 8% and 6% of the total costs, respectively. The KCR, at least during the annual period studied, expended a significant portion of its resources on training staff at other registries, especially GCR staff. This training is reflected as a cost to the KCR, when in fact the training benefits other registries.

Fig. 3 provides the details on spending by registry activity categorized into the broad components of fixed cost, variable core cost, and variable other cost activities. These detailed activity-based costs take into account the percentage of staff time spent on each activity. Overall, the highest cost activities were administration and indirect costs, followed by data collection and abstraction. Among the fixed cost activities, administration and management made up 83% of the cost, and training of registry staff made up another 8% (not shown in figure). Within the variable core cost category, 78% of the cost was for data collection, abstraction, and entry into the registry database (not shown in figure). Case consolidation, quality assurance, and sharing of cases accounted for the remaining variable core costs. The KCR performed several variable other cost activities. The highest cost was allocated to training of others by the registry (which is overall the third highest cost activity). Other activities included publications, activity follow up, and research studies. In terms of cost per cancer case, fixed costs were 11,736 Ugandan Shillings (UGX), variable core costs were 11,303 UGX, and variable other costs were 5162 UGX (4.25, 4.09, and 1.87 respectively in 2014 US dollars). The total cost of the KCR during 2014 was about 53.7 million UGX, which was equivalent to a little less than 20,000 US dollars in 2014.

The GCR is a start-up registry; however, as shown in Table 3, the registry will likely require stable and continued funding support to sustain operations in the future, as current funding is from a research study. Covering a larger geographic area may increase costs for the GCR; the staff need to travel to several other major hospitals in the coverage area, and the furthest hospital requires a trip of some 100 kilometers. Overall, we anticipate that the cost of equipment and administration may remain relatively the same for both registries. Currently, all the GCR start-up operations are supported by the host institution with assistance from international agencies, such as the IARC, in terms of technical expertise as well as free cancer registry software.