There are currently 2 million cancer survivors in the UK, which is estimated to increase to 4 million by 2030 (
Maddams et al, 2009). Some survivors experience ongoing physical and psychosocial difficulties, including problems with fatigue, emotional health and work, yet many survivors report comparable quality of life to their general population peers (
Arndt et al, 2005;
Helgesson et al, 2007;
Short et al, 2008;
Hoffman et al, 2009;
Harrison et al, 2011). Current understanding of who experiences particular survivorship difficulties and when, however, is limited; research to date has mostly examined a narrow range of outcomes in small-scale, short-term, non-UK studies (
Bloom et al, 2007;
Foster et al, 2009). A priority of the UK National Cancer Survivorship Initiative (NCSI) is to improve the understanding of survivors' experiences through increased collection of patient-reported outcome measures (PROMs, i.e., questionnaires), across multiple quality-of-life domains for many years post diagnosis. Understanding of survivorship issues is vital to inform service development and interventions, and facilitate targeted provision, with the latter being increasingly important as the number of survivors rises in a difficult financial environment.
Prospective longitudinal PROMs collection is expensive, owing to the high administrative and staff costs of repeated questionnaire administration, and ongoing patient tracking and communication. Moreover, PROMs collection hinges on high rates of sustained patient participation, yet poor accrual and high attrition are common problems in oncology and research generally (
Easterbrook and Matthews, 1992;
Dilts and Sandler, 2006). There is potential to reduce the cost of PROMs collection, by making many of the involved processes electronic and/or Internet-based. Administering PROMs online eliminates the paper-based costs of printing, distribution, data entry and physical storage, and is therefore estimated to be cheaper than paper administration by over 75% (
Greenlaw and Brown-Welty, 2009;
Russell et al, 2010). Automating monitoring of patients' questionnaire activity, and preparation of related communications, through use of a tracking database, can greatly reduce staff costs. Similarly, Email communications avoid the expense of postal mail. Online PROMs may also improve patient convenience, and thereby recruitment and retention. Internet questionnaires can be immediately accessed, completed and submitted, 24
h a day, year-round, from numerous web-accessible devices and locations. Studies that have assessed and established psychometric equivalence of electronic and paper PROMs have also reported finding a patient preference in favour of Internet
vs paper administration (
Handa et al, 2008;
Bishop et al, 2010).
To be most informative, PROMs must be linked and analysed with patients' diagnostic and treatment information; this is important to identify clinical predictors of survivorship difficulties and enable risk stratification. However, clinical data linkage can be logistically and ethically complex, because of issues such as data completeness and security, and also time consuming and costly (
Bohensky et al, 2010). A review by
Lipscomb et al (2007) highlighted failure to link PROMs with clinical data as a common problem in cancer survey research, and urged the establishment of ‘data systems that incorporate not only traditional clinical and epidemiological parameters, but patient-reported outcomes' (p 292). The UK has a comprehensive network of cancer registries that collate, link and store clinical information on all cancer patients to produce incidence and survival statistics; the network comprises eight regional English registries and three national registries in each of Scotland, Wales and Northern Ireland. English data are collated by the National Cancer Intelligence Network (NCIN) in the National Cancer Data Repository (NCDR), which is centrally linked with other data sets including Hospital Episode Statistics (HES). Data linking through the registries would be logistically efficient and enable PROMs to be linked with a standard clinical data set. Registry linkage would also capitalise on costly work already undertaken to collate and validate patients' clinical data, and, as registries work to the highest standards of data protection and security, would ensure compliance with all relevant ethical, NHS and legal mandates. By means of the NCDR, registry linkage also affords potential for linking (English) PROMs to other data such as HES.
We aim to facilitate epidemiological measurement of UK survivors' experiences by developing a technical system for regularly collecting PROMs and linking to clinical information: the electronic Patient-reported Outcomes from Cancer Survivors (ePOCS) system. Specifically, we aim to develop a system in which PROMs are completed online and stored with clinical data in the registries/NCDR, and in which patient tracking is electronically automated and communication is Email-based. As the Internet and registry network both provide enduring UK-wide coverage, such a system is potentially sustainable and UK scalable. Recent years have seen rapid growth in the use of electronic PROMs and analyses of linked data sets (
Gwaltney et al, 2008;
Snyder et al, 2009;
Bohensky et al, 2010). In The Netherlands, the PROFILES Group (Patient Reported Outcomes Following Initial treatment and Long-term Evaluation of Survivorship) have very recently set up a ‘web-based registry' to facilitate cancer PROMs collection, and which links to clinical data in a regional cancer registry (
van de Poll-Franse et al, 2011). In the UK, however, there is currently no system for regularly collecting cancer PROMs and linking to registry data.
Patients must be recruited for PROMs-based research (i.e., identified eligible and provide consent). Any PROMs collection system must be able to receive information of newly recruited patients, and must be developed, therefore, with a view to the context(s) of recruitment. It is not permissible to contact patients directly via UK cancer registries. Research suggests that patient recruitment rates are generally higher in secondary and tertiary hospitals than primary care general practitioner practices (
Hunt et al, 2001;
Hetherton et al, 2004;
Hummers-Pradier et al, 2008). We anticipate that recruitment will be more effective and sustainable if undertaken by clinical teams than by non-NHS research personnel. Therefore, we decided to develop (and test) ePOCS within a context of hospital-based recruitment led by research nurses (RNs).
This article describes the technical development of the ePOCS system, which entailed designing and building new constituent technical components, and establishing and adjusting component-connecting data flows. The system is currently being tested in a feasibility study with non-metastatic breast, colorectal and prostate cancer patients in two Yorkshire NHS Trusts, and using the Northern and Yorkshire Cancer Registry and Information Service (NYCRIS). Feasibility evaluation will be the subject of future papers (e.g., patient participation rates, reliability of the informatics, running costs). In this paper, we aim to (1) delineate the methods and processes involved in designing and building the system; (2) describe the developed system in terms of its technical components, data flows and involved human activities; and (3) discuss the challenges of development, and those potentially involved in sustaining and scaling up the system.
This article has been 
![[less-than-or-eq, slant]](/corehtml/pmc/pmcents/les.gif)
6, 9 and 15 months post diagnosis). In the feasibility study to date (November 2010 to August 2011), over 550 patients have joined the ePOCS system (≈60% consent rate), most have provided an Email address for communications (>80%) and ≈75% have fully completed PROMs at the first timepoint (80 questions total); the proportion of missing responses is <1% (i.e., patients' responded ‘prefer not to reply'), and <2% of patients have withdrawn consent. Few patients have contacted the administrator team for help (<10%); moreover, many queries have concerned the same easily rectifiable issue (misreading zeros and letter Os in ID/password). Thus far, the most common reasons offered for not joining ePOCS are a lack of interest in and/or access to the Internet.