Background
Patient reported outcome measures (PROMs) are designed to measure patients’ views of their symptoms, own functional status, treatment satisfaction and health related quality of life in relation to specific disease or conditions [
1‐
4]. PROMs are an important measure of patients’ perspective of care outcomes as they provide insight into the impact of a disease and its treatment on daily lives. PROMs can assist clinicians to work with patients to achieve a level of care that meets their needs; this has been demonstrated to improve patient-provider communication [
5].
PROMs may be self-administered or administered by another person (third-party). Instruments used to collect PROMs should be validated for the mode with which they are being administered. Self-administration may include surveys that patients complete on paper or electronically (e.g. via links provided in an email address to an online form or through Applications (apps) that patients can download. Tools administered by another person may include those completed on paper or electronically with assistance or those administered over the telephone [
6,
7]. With the increasing number of Internet users, greater opportunities exist for collecting data, through mechanisms such as email and web-based surveys [
8,
9]. An emerging trend in health-related survey research is the use of a mixed-mode approach. In the mixed-mode method, individuals may respond using a single or combination of different modes, such as only telephone or mail followed by telephone [
10]. Using a mixed-mode approach compensates for the weaknesses of each individual mode at affordable cost [
11]. Survey mode, length and content of the survey and incentives will impact the response rates and the cost of data collection [
12].
The amount of clinical data being collected is growing exponentially; largely due to computer-based information systems [
13]. In Australia, the number of known registries collecting clinical data has risen from 28 in 2006-07 to 37 in 2012 [
14]. Clinical quality registries have received increasing attention as a means of improving quality and reducing the cost of health and medical care, through identifying variations in clinical practice and care, and assessing the uptake of effective treatment [
15,
16]. A number of clinical quality registries collect PROMs and provide reports on outcomes to hospitals. Examples can be found in trauma [
17], joint replacement [
18] and renal disease [
19]. PROMs are being developed by the American Society of Clinical Oncology to benchmark hospitals in relation to symptoms and functional status following cancer treatment [
20].
For any individual research study, the mode of data collection is influenced by time, available resources and the population being targeted [
21]. A number of studies have compared response rates using different modes of data collection [
22]. A meta-analysis published in 2009 found that email surveys have lower response rates compared with mail surveys (20% vs 53% respectively) suggesting that, despite rapid growth of information technology, mail surveys appear to be superior to email in collecting survey data [
22]. High response rates have been obtained when follow up attempts are intense and personalised. For example Steineck et all reported very high response rates of 89 to 99% across multiple time periods by following a regimen which included an introduction letter and a telephone call to establish contact prior to a survey being posted and a “thank you and reminder” card following return of the survey [
23,
24]. It is unclear whether such a labour-intensive approach is sustainable at a population level.
Response rates have been found to vary among study populations. Postal surveys with three reminders have shown demonstrably better response rates among general practitioners compared with a telephone survey [
25]. A randomised control trial (RCT) of junior medical staff and faculty members comparing electronic and postal surveys found that response rates were similar, but the average response time for electronic surveys was shorter for the residents’ group compared with the faculty group (3.8 days vs 8.4 days, p < 0.001) [
26].
A recent meta-analysis by Rutherford et al. [
27] investigated whether the mode of PROMs administration introduced bias into the patient reported outcome results. Findings suggested that there was no bias associated with whether PROMs were collected electronically (computer including web, touch screen, hand-held device, video conference, computer assisted telephone interview), via paper self-completion (hard copy) or via assisted completion in clinics or home. The authors of the study recommended further research using experimental designs to measure the mediators of mode effects on data quality, measurement equivalence, reliability of assessment for individuals and the impact of setting and combination of data collection method over time [
27].
A cost-effectiveness study by Sinclair et al. (2012) found that postal survey costs were lower compared to both internet and telephone. The cost of a completed response using a personalised postal survey (24.75 Australian Dollars) was slightly higher than the generic postal survey, a generic internet survey and a personalised internet survey cost was almost double of a personalised postal survey and a telephone survey cost was highest among all methods [
28]. Another study conducted by the Australian national stroke registry, found that telephone follow up for patient with acute stroke or transient ischemic attack was more expensive but more effective in terms of completion rates than follow-up by postal mail [
16].
The Victorian Prostate Cancer Registry (PCOR-VIC) was established in 2009 to monitor treatment and outcomes of men diagnosed with prostate cancer in Victoria. PROMs are collected to assess the impact of prostate cancer diagnosis/treatment on urinary, bowel, hormonal, and sexual function and bother using the EPIC-26 survey [
29]. The EPIC-26 survey has been validated for telephone and self-administered survey (paper or online) and is currently only administered by telephone [
30]. The response rate has varied over time as modifications have been made to the registry and is currently at 85%. Alternative methods of PROMs administration have not been systematically assessed for their cost-effectiveness and feasibility. Although previous studies demonstrated lower response rates and in some cases increased costs of surveys delivered electronically, these were conducted several years ago and on a different population. As the PCOR-VIC is now contributing to a newly developed Prostate Cancer Registry-Australia and New Zealand, [
31] the aim of this project was to assess the most cost-effective approach for collecting PROMs in a prostate cancer population.
The current study protocol describes the design of an equivalence RCT to assess the cost-effectiveness of three different methods of PROMs data collection using the EPIC-26 survey for patients diagnosed with prostate cancer.
The primary objective of the trial is to compare the completeness of survey data obtained using the three different data collection approaches for reporting on PROMs.
The secondary objectives are to:
a.
Estimate recurrent costs of data collection using telephone, postal services/mail and electronic mail (email) for PROMs data in PCOR-VIC.
b.
Compare the cost-effectiveness of the three different methods of data collection.
c.
Develop a cost projection model to estimate the cost for nation-wide scale-up of administering the PROMs data collection tool in the most efficient setting for follow up of prostate cancer patients in Australia.
Methods/design
Setting
Men who are diagnosed with prostate cancer in Victoria, contributing to the PCOR-VIC and who are interviewed by researchers to collect PROMS, will be invited to participate in this study. Since its establishment in 2009 the registry has expanded to 33 hospitals across the state, representing approximately 75% of the Victorian population [
32]. Men are eligible for inclusion on the register if they have had a histologically confirmed diagnosis of prostate cancer that is notified to the Victorian Cancer Registry by the hospital.
Trial design
The study design proposed for this evaluation is an equivalence RCT design. Participants will be individually randomized to one of three independent groups receiving the PROMs instrument by email, post or by telephone. Due to the nature of the intervention, it is not possible to blind the researchers or study participants.
To collect costing data we will use an Activity Based Costing (ABC) method [
33] and structured questionnaires to estimate the cost of the operational activities of the three different methods of follow up. The ABC method is useful for understanding key activities of any programs and interventions and allows identification of (i) implementation levels and composition of costs; (ii) variations in how an intervention is implemented over time and associated cost implications; and (iii) resulting costs of increasing coverage of cost-effective data collection methods. This method is flexible, so its resulting estimates can be easily understood and adapted to measure the cost of data collection of the three different methods. Costs that will be considered include personnel cost, cost of supplies (e.g., envelopes, printing etc.), cost of training of data collectors and cost of operation and maintenance (e.g., telephone bill, internet bill, rent etc.) for each of the data collection methods.
Recruitment of patients
Recruitment of patients to the PCOR-VIC has been previously described [
34]. In summary, patients diagnosed in recruiting hospitals and notified by the hospital to the registry are sent details of the registry in an explanatory statement by mail. In the explanatory statement details on what data will be collected from the patient’s medical record and directly from the patient and how a man can opt off the registry if he chooses not to participate are included. The explanatory statement also contains the contact details of both the hospital where the patient was diagnosed and the university conducting the research and hosting the registry. Clinical data are collected on men who do not opt out. A waiver of consent enables clinical details to be collected from men who have died after diagnosis.
Eligible men are contacted by centrally-located university call-centre follow up staff to confirm that clinical data are accurate and up-to-date. For data collection contact is made any time within a window period of 21 days on either side of the anniversary date for data collection (henceforth recorded as the “Anniversary Date”). This is 12 months from the date of the positive biopsy for patients who do not proceed to active treatment or only receive androgen deprivation therapy; or 12 months from the date on which final initial treatment, or course of treatment finished. This is also the date for surgery and low-dose rate (seed) brachytherapy procedure. For radiotherapy and chemotherapy this is the date that the last dose of therapy was provided as well.
Inclusion criteria
Men will be included in this study if they are eligible and have been included on the PCOR-VIC, are aged >18 years and answer the telephone when contacted by data collectors in the 21 days leading up to and including their Anniversary Date.
Exclusion criteria
Men will be ineligible for inclusion in the RCT if they opt off the registry, have died in the period between being recruited to the registry and telephoned, do not speak English, are identified as being hearing or mentally impaired when contacted by data collectors to administer the PROMs, have been diagnosed by Transurethral Resection of the Prostate (TURP) and their treating doctor has requested that we do not contact them for follow up, or if they answer the telephone after their Anniversary Date.
Acknowledgements
We would like to thank Dr Arul Earnest, Associate Professor, Department of Epidemiology and Preventive Medicine School of Public Health and Preventive Medicine of Monash University for assisting us in calculating sample size and generating random numbers. We also would like to thank Dr Nupur Nag, Research Fellow, Department of Epidemiology and Preventive Medicine School of Public Health and Preventive Medicine of Monash University, for assistance in developing the graphics.