Cancer Control Planning: Supplying the Evidence

Most advances in clinical practice are the result of good quality published research. However, over-reliance upon outcomes of clinical studies conducted in high income countries may be inappropriate when developing guidelines or planning cancer control strategies in poorer resourced countries. Guideline developers and cancer control planners need to be able to identify, and have access to reports of good quality research conducted in their own, or similar countries, in order to be in a position to plan and conduct feasible interventions.

Since 2007 INCTR has developed two databases as part of a demonstration project that could lead eventually to a Global Database for Cancer Control in Developing Countries, the construction and analysis of which will provide both negative (relating to poor quality information and gaps in the evidence base) as well as positive (meaningful data that can be used to plan interventions) information. The first database (previously reported in the winter 2007-2008 issue of Network [Vol 7 issue 4]) is the result of a collaborative pilot project involving INCTR UK (Mark Lodge, Biying Liu) and NCI Cairo (Prof. Hussein Khaled, Dr Atef Badran, Mrs Howaida Ali Alsayad, Ms Shaymaa Badran). Support was provided by the European School of Oncology, the European Society of Medical Oncology and the Egyptian Foundation for Cancer Research. This project was designed to identify published reports of Egyptian research in five areas of cancer care: breast cancer, bladder cancer, lymphoma, childhood cancers and palliative care. The second database has been created by INCTR UK from studies identified by Mark Lodge, Biying Liu (Oxford UK) and Dr Marilys Corbex (Cairo, Egypt) as a result of a commission from the National Cancer Institute Maryland USA to catalogue published articles of special relevance to breast cancer control in low and middle income (LMI) countries and to summarise the body of research evidence in terms of its scope and quality. Sufficiently high quality material will be used to create systematic reviews capable of being answered by the available evidence.

Both databases have been built using the same basic strategy. Clear Inclusion and exclusion criteria were established: all studies relevant to a particular cancer and either Egypt or LMI countries published between 2000 and 2008 were included; animal studies were excluded. Individual sensitive search strategies were developed for multiple databases (including Medline, Embase, Scopus, Biological Abstracts, Popline) and conference proceedings (ASCO, ESMO, UICC, INCTR and San Antonio Breast Cancer Symposia). Other sources of studies were contacted and invited to contribute. In the Egyptian project great pains were taken to search the contents pages of over forty non-indexed Egyptian journals: a strategy only made possible by the willing collaboration of many senior university librarians across the country. Particular acknowledgement should also be paid to Dr. Don Ekwueme for generously sharing important data from a parallel breast cancer literature search conducted on behalf of the Centers for Disease Control Foundation and Susan G. Komen for the Cure Foundation. The titles and abstracts (where available online) were downloaded or entered by hand into a ProCite 5 database, de-duplicated and coded by area of care (e.g. prevention, detection, diagnosis, treatment etc.).

In the Egyptian project, 1383 studies met the inclusion criteria. The results of the searches are presented in Table 1. (NB: Some articles are necessarily coded more than once; e.g. ‘Breast cancer’ + ‘Palliative care’.)

	Database searches	Non-indexed studies	Total
Breast	215	217	432
Bladder cancer	262	157	419
Lymphoma	99	234	333
Childhood cancer	73	105	178
Palliative Care	54	30	84
Table 1. INCTR/ESO Egypt study: Indexed and non-indexed studies identified through electronic (database searches) and manual searches (non-indexed studies).

In the second project, 4871 articles relevant to breast cancer control in LMI countries were identified, including 219 reports of randomized controlled trials (RCTs). Countries producing the most research, collated by WHO Region, and ‘Global’ reports are presented in Table 2. (NB: Since reports of multicentre studies will be represented more than once if they involved collaboration across WHO Regional boundaries the figures presented in Table 2 will not total to 4871 (100%).)

WHO Region	No. of Studies (%)	Leading countries (n)
Africa	230 (5%)	South Africa (80)
Americas	726 (15%)	Brazil (330)
South East Asia	569 (12%)	India (451)
Europe	1465 (30%)	Turkey (674)
Eastern Mediterranean	590 (12%)	Egypt (190)
Western Pacific	1230 (24%)	Mainland China (777)
‘Global’	91 (2%)	USA (46)
Table 2. Breast cancer in LMI countries: No. of publications and "leading" countries arranged by WHO Region.

Both literature searches demonstrate that a simple search on Medline or even two databases is insufficient to identify all studies relevant to LMI countries and that establishing a comprehensive evidence base currently requires multiple strategies. Of the 1383 studies identified in the Egyptian project 675 (49%) were identified by electronic searches of four databases and 708 (51%) were only identifiable by searching the content pages and local online citations of Egyptian journals. Of the 4871 breast cancer studies identified only 2499 (51%) were indexed on Medline; 1096 (22%) were available on Embase only; 702 (15%) were not found on either database but were retrieved from non-indexed conference proceedings. (These latter percentages would undoubtedly have been lower if hand searches of un-indexed journals published in low or middle income countries for breast cancer articles had been undertaken.)

Details of the studies identified by the ESO/INCTR Pilot Study will be accessible next year on the University of Cairo website (http://www.cu.edu.eg). Bibliographic citations of the breast cancer articles will be made accessible through the new Web-based BHGI/INCTR Breast Cancer Control Library being developed in collaboration with INCTR by Breast Health Global Initiative (BHGI) staff at the Fred Hutchinson Cancer Research Centre, Seattle WA USA.

Bibliographic databases of published studies on their own will be of limited use to decision makers: their most important function is to pull together the raw material from which evidence-based cancer control strategies may be developed. Faced with such a large body of evidence, there is a clear need for taxonomic surveys to arrange and synthesise the data into useful packages, quality assessment of, for example level 1 evidence (randomized controlled clinical trials) and systematic reviews of high quality information in order to facilitate cancer control planning and research agenda-setting in poorer resourced countries.

In the high income countries the technological advances in information management of the last thirty years, coupled with the growing awareness of the dangers arising from methodological bias in the reporting and use of research outcomes, have transformed how research agendas have been developed. In particular the attention paid by Governmental decision-makers (e.g. the UK Department of Health) to the importance of evidence from research syntheses stems in part from the recognition that advances in information technology (IT) now allow us access to more scientific evidence than ever before both through increasingly comprehensive data bases and also through linkages that permit the researcher to progress along a specific desired path in collecting all relevant material. Several major projects, such as the US National Cancer Institute’s caBIG and the UK National Cancer Research Institute’s ONIX permit simultaneous searching of multiple, integrated data bases containing different but complementary information, such as molecular profiling and clinical characteristics, and subsequent analysis pertinent to a specific question raised by a researcher uninvolved with the data collection.

“All existing sources of evidence, especially systematic reviews, must be considered carefully before undertaking research. Research which duplicates other work unnecessarily, or which is not of sufficient quality to contribute something useful to existing knowledge, is unethical.” Research Governance Framework for Health and Social Care Department of Health 2nd ed. April 2005 2.3.1

The model established by Cochrane systematic reviews – that each review should conclude with a brief summary of the review’s implications for clinical practice and future research – has become accepted as a methodological benchmark. The contribution of properly conducted research syntheses to improvements in clinical treatment is universally acknowledged with systematic reviews and meta-analyses now occupying the apex of most research hierarchy pyramids (sometimes vying for top position with large randomised trials). The equally valuable role systematic reviews have to play in determining research agendas is less widely recognised. This role (as outlined in the five research functions of systematic reviews listed in Box 2) can be summarised as helping facilitate the preparation of high quality study protocols for ethical review and protecting patients from recruitment into badly designed or unnecessary studies.

Whereas the original data INCTR has identified comes from studies conducted in LMI countries, expertise in preparing systematic literature review and meta-analysis still resides mostly in high income countries. Out of the 4871 published articles identified in the INCTR Breast cancer literature search only 29 systematic reviews and meta-analyses (<0.6%) were identified. Therefore, in parallel with improving access to published reports of research studies, capacity in research synthesis expertise needs to be built in low and middle income countries.

Ad hoc introductory programs are useful in raising awareness but can be expensive and have not been shown to be very productive in terms of published output. The best way to build capacity is to ensure that knowledge of systematic review methodology - how systematic reviews are prepared and what their limitations are - becomes an integral part of the postgraduate medical training (PGMT) curriculum. Universities and other providers of PGMT have a responsibility to require that future generations of health care professionals are fluent in the fundamentals of evidence-based practice. However, they may also need to build their own bibliographies, since it is clear that the well-known data bases fail to incorporate evidence from a large number of journals. Indeed, as countries increase their research outputs (to say nothing of practice guidelines and information for the public) individual data bases may not be able to handle the deluge of data, such that simultaneous searching of multiple data bases scattered around the world will be required, along with applications that permit rapid manipulation of both bibliographic and other forms of data. The integration of hospital based data, bearing in mind issues of confidentiality, may be a future step in a holistic approach to data gathering and use. Such comprehensive approaches will require national health systems, international collaboration and the development of robust interfaces to ensure interoperability. Indeed, such approaches will permit the use of a considerable amount of data collected outside the research arena, which is presently largely inaccessible and therefore wasted, to be used to the benefit of populations and patients.

Five Research Functions of Systematic Reviews: Identify all relevant published studies Provide methodologically transparent summaries/syntheses of the scope, outcomes and quality of published research Prevent wasteful duplication of research effort and resources Identify areas where new or better research is needed Identify methodological challenges previously encountered by authors

Table 3. Five Research Functions of Systematic Reviews.

Five conclusions may be drawn from the progress of the Global database project so far:

1. Research articles originating from low or middle income countries are under-represented in the more accessible indexed databases.
   
2. Searching one or even two reputable databases is insufficient to identify all relevant studies on an issue.
   
3. The development of guidelines and national cancer control plans can be facilitated through building an accessible and maintained database of relevant research literature.
   
4. More systematic reviews of reports of studies conducted in low and middle income countries are needed, but will only be possible after all data is catalogued, made widely accessible, and assessed for quality.
   
5. Postgraduate medical training curricula should include an appreciation of systematic review methodology in order to encourage the development of evidence-based clinical practice and research.