Using technology to deliver cancer follow-up: a systematic review

The population of interest was adults with cancer. The intervention was cancer follow up using technology and the control usual care. Inclusion criteria were randomised controlled studies published in English between 2000 and 2014, whose intervention included a telemedicine or telehealthcare element in the intervention. Studies not meeting these criteria were excluded. The key outcomes of interest were patient acceptability (satisfaction), clinical safety and cost effectiveness.

A search strategy based on key words to reflect the review aim was designed in conjunction with a medical librarian and is included as Additional file 1.

The searches were run in February 2014 on the following databases: All EBM Reviews; Embase; Medline (No Revisions); Medline (Non-Indexed Citations); CAB Abstracts. Retrieved citations were exported to Refworks (http://​www.​refworks.​com). All identified titles were read and those not meeting the inclusion criteria were excluded, as were duplicates. Abstracts of the remaining studies were screened against the inclusion criteria and full articles were then retrieved.

Critical appraisal of selected studies was undertaken using a standardized checklist [14]. This was done by two researchers (RD and SH). There was subsequent discussion of assigned scores, and discussion and resolution of any differences by consensus. Papers were analyzed thematically considering particularly outcomes that related to themes highlighted in the review question, namely, clinical safety, patient acceptability, cost effectiveness and impact on quality of life. Clinical safety was defined as any outcome related to recurrent cancer or mortality. Patient acceptability was defined as outcomes reflecting how easily patients had found engaging with an intervention, the extent to which it met their healthcare needs, or how it impacted upon their satisfaction with services. Cost effectiveness related to the reporting of appropriate economic data. Quality of life related to any outcomes reflecting symptoms or accepted or validated measures of health related quality of life. Information was collected on the purpose, process, results and limitations of each study using a standardised data collection template. All outcomes were considered with particular attention paid to issues of patient acceptability and satisfaction, clinical safety, cost and impact on quality of life. Where quantitative data was presented it was tabulated as p values, confidence intervals and effect sizes. A narrative analysis of all papers was also conducted to identify the emergent common and contrasting themes from the reported studies.

Figure 1, a PRISMA diagram, displays the data for the number of titles initially identified, then excluded along with duplicates and the final number of randomised studies identified and included.

Seventeen papers pertaining to 13 randomised studies were included in the review. Tables 1 and 2 considers and displays each of these studies in detail including terms of patient population, intervention, outcome measures, key results, study quality score, measures used and detailed results including effect sizes.

A standard checklist was used to critically appraise the included studies using standard criteria [14]. Scores are shown in Table 1 and ranged from 17 to 23. One study attracted a quality score of 6/10, mainly due to poor expression of the study aims and description of the participants [15]. Thus, these studies were generally of good quality and reported to a high standard. In all cases the researchers had addressed issues of internal validity and recognised the issues created by not being able, in most, cases to blind participants to the intervention. Most studies failed to employ blinded data collection. In all studies there was good recognition of, and controlling of, potential confounding leading to high scores in this domain. Most of the studies were, however, underpowered with respect to clinical outcomes. The main issue of quality with the study was a failure to demonstrate external validity. Only four of the included studies included information on non-respondents, generally reporting that they were older and of lower educational status [11, 17, 18, 20]. This underscores the fact that technology trials as currently reported exhibit considerable recruitment bias. This was further illustrated by some of the studies not reporting on non-respondents acknowledging that this was a limitation and highlighting the fact that they had, for example, recruited only urban or white participants [15, 18, 28].

The limited evidence available suggests that using technology for cancer follow-up is acceptable to patients and clinically safe. However, there is currently insufficient evidence to definitively state whether or not remote cancer follow-up using technology is likely to be cost effective, since most existing randomised trial evidence relates to alternative models of follow-up using standard telephone calls only. The results are important as a prompt and guide to future research since, in the face an ageing population and increasing number of cancer survivors, the need for safe, acceptable and economically viable alternatives to current resource intensive cancer follow-up care delivery models is urgent. Existing evidence suggests that scheduled cancer follow-up can employ elements of technology without reducing patient satisfaction, compromising safety, impairing quality of life or increasing psychological distress.

As far as we aware this is the first systematic review to focus on the use of technology to deliver cancer follow-up. The methods of the review were thorough and robust and the authors have confidence that the search strategy was suitably inclusive to identify most relevant studies. We chose to include only randomised controlled trials in the review and did not include other studies types, such as interrupted time series studies. This was principally due to our concern that trials of new technologies are particularly prone to selection bias (i.e. the recruitment of enthusiasts). Whilst the randomised trial is not a panacea to this problem in technological research we believe that by restricting our review to a method which explicitly addresses this issue was the best approach. A further issue is that the CONSORT statement which promotes quality in the reporting has been widely accepted and implemented. Furthermore randomised trials, in general, include harder outcome. We believe our results have borne out our approach.

The review was limited to studies written in English. Thus it is possible that some studies conducted in other parts of the world were missed. This is particularly so when it is considered that many non-English speaking countries, for example, Germany are key-players in technological innovation. We do think, however, that since English is practically the “lingua franca” of medical research it is unlikely that we have missed any major randomised trial by employing this limitation. It does seem most likely that these would have been published in a journal employing the English language for which the search strategy was exhaustive. Further support for this view can be obtained from the fact that both trials currently registered on the ISRCTN register of using technology in cancer follow-up are in non-English speaking countries but fully described and registered in English.

The majority of existing studies were conducted in the developed world in women with breast cancer, a group which will potentially represent the more enabled and affluent end of the cancer survivorship spectrum. This limits the extension of the current findings to other settings, cancers and patient groups, emphasising the need for further research. It could also be that there are patient groups in whom it would be inappropriate to attempt to utilise technology in their follow-up, an issue that the current evidence goes no way toward identifying or addressing. Furthermore, there was insufficiently consistent reporting within the studies of inclusion criteria with respect to time since completion of treatment or stage at diagnosis to draw meaningful conclusions with respect to these parameters. A further limitation has occurred in the synthesis and interpretation of the evidence. There were surprisingly few randomised studies identified and the majority used telephone follow-up only, with only a few adopting more recent technological innovations such as PDAs. Thus, given the increasing potential of modern digital technologies the included interventions were relatively simple and low-tech. Furthermore, the way in which technology could enable patients to send information to healthcare professionals has not yet been meaningfully explored. Although the interventions addressed several aspects of follow-up care (recurrence monitoring, monitoring treatment effect, information provision) none were comprehensive, and most used technology to deliver one aspect of follow-up care as an adjunct to more traditional follow-up models. Because of this direct comparison of interventions was extremely difficult given the different technologies, interventions and outcomes reported. Nevertheless, we believe the current review provides good evidence of the potential of technology to deliver safe effective cancer follow-up in the future. The review also highlights the need for future trials to seek to provide definitive information on patient acceptability, clinical safety and cost effectiveness.

There is a growing recognition that cancer follow-up services are increasingly stretched by a growing population of cancer survivors [1, 3, 6]. This has prompted researchers to develop alternative models of care which have utilised different healthcare professionals in care delivery and/or shifted the focus of care from hospital to the community [4, 29, 30]. Despite, this, these models will probably continue to consume a large and growing amount of healthcare resource and still inconvenience certain patient groups. The randomised studies included within this review are striking in the respect that they demonstrate that few mature complex interventions utilising technology, other than standard telephone calls, have been developed to a standard sufficient for subjection to a randomised controlled trial. The need to do this is, perhaps, one of the key implications of our work. There is an emerging vision that modern technologies, evidenced by policy initiative worldwide, could potentially enhance current models or offer other alternatives to the future delivery of high quality cancer care [31‐38]. There is considerable evidence from other disease areas of using modern technologies to good effect [39]. For example, text messaging has been found to increase self-efficacy and adherence to medication in young people with diabetes and has been proven to be an effective aid to smoking cessation [40, 41]. Online systems which enable patients to access their own health records have enjoyed some success as a means to educate patients and assist in their self-management, for example, in the USA, Kaiser Permanente’s My Health Manager has demonstrated significant reductions in primary care contacts [42]. The CHAMPION project has enabled disabled adults to improve communication with healthcare professionals by inputting information on their own care needs into a database, for example, uploading a video to demonstrate to clinical staff how they use their communication devices [43]. However, there is now a striking need to develop high quality and innovative technological interventions to support growing numbers of people with cancer.

Inspection of the International Standard Randomised Controlled Trial Number Register indicates that increasing numbers of clinical trials of technological applications to healthcare are occurring [44]. The register lists twenty-three active trials on telehealth, telemedicine or ehealth, including trials in men with prostate cancer and another amongst patients with breast or colorectal cancer [45, 46]. Nevertheless, this is surprisingly few randomised trials given the explosion in technological innovation in recent years. It could be that technology is evolving so fast that potential innovative technological interventions become outdated before they can mature sufficiently to be subjected to randomised trials.