Short summary

Survey data from the three completed surveys undertaken at REHPA: DANCAS (Cardiac Arrest, 2021), NUTRI-HAB (head and neck cancer, 2019) and PReTT (Parkinson's disease, 2020) will be coupled with data from Danish national registries on vital status, socioeconomic characteristics and co-morbidity. This provides a unique opportunity to assess patterns in response/non-response and to explore the cost-benefit of using postal surveys in addition to online electronic surveys.

Rationale

Health surveys and patient-reported questionnaires have become an increasingly important method of collecting diverse health information from the general population and people with life limiting or life threatening conditions. Surveys reveal information not available in national health registries, including participants' symptoms, behaviours, knowledge or opinions. However, the validity of survey data and the conclusions drawn from them can be gravely compromised if respondents are not missing at random.

Reasons for non-response are difficult to determine but have been associated with participants' socioeconomic characteristics or the mode of survey delivery. Electronic modes have become an increasingly popular method of conducting survey studies. They are cheaper to administer with no print or postage costs, and results are instantly available to researchers. In Denmark all citizens are offered an e-Boks account when they turn 15 years old. All public sector correspondence is sent electronically to a person's e-Boks. People with no access to a computer, who have a functional impairment, are homeless or have linguistic difficulties, can be exempted from e-Boks. If questionnaires are only sent electronically the non-e-Boks group is excluded from any study population thereby producing potential bias in the results. Given these advantages and challenges, it is important to understand the effect electronic-only surveys may have on survey response rates and whether it may lead to unintentional response bias.

A previous study, using the Danish general population from 2015-2016, showed almost no differences in socioeconomic characteristics between respondents and non-respondents between an all-postal survey or electronic survey + postal (postal was only for people without e-Boks). They also found the cost of providing postal surveys was ten times the cost of the electronic survey. However, 'electronic only' versus 'electronic+postal' has not been similarly investigated and results may be different for participants with health conditions. A study with colorectal cancer patients, who tend to be older than the mean age of the general population, found response rates did not differ when participants were offered electronic or postal formats but importantly, the older patients (≥70 years of age) preferred to complete surveys by post.6 In contrast, a survey of Myeloproliferative neoplasms patients from 2013 showed that the highest response rate was from patients 70-79 years old, living with someone, of a Danish/Western ethnicity and with a higher level of education. Indicating that socioeconomic characteristics may have a different effect on survey response depending on the health condition.

Age may not be the only socioeconomic factor associated with a lower response rate. Gender, civil status, living alone, education level, income and immigrant status have been shown to be associated with variances in co-morbidities, referral to health services and survival rates within difference health conditions. Hence, these socioeconomic characteristics may also be associated with differing response rates in health surveys. In addition, co-morbidity burden may affect participants' ability or willingness to respond to surveys. Fatigue, pain, depression and anxiety are all known to affect concentration and memory and cognitive problems are often inherent to neurological conditions.

Therefore, a high co-morbidity burden may reduce participants' use of electronic contact methods, for example, email or e-Boks, or mean they have difficulty navigating and completing electronic surveys. The argument for 'electronic only' surveys may not be as simple as ensuring that participant socioeconomic characteristics are the same as 'electronic+paper', and so we will also investigate whether co-morbidity burden effects response bias.

Understanding whether systematic response biases exist, and the effect of health condition on delivery mode is important to ensure survey respondents represent the target population as much as possible. Further, given the differences in symptoms and disease trajectory in different health conditions, survey format response bias may occur in some patient groups but not others. Finally, given the very high and increasing costs of sending postal surveys it is essential funders and researchers understand the cost versus benefits of using postal surveys in order to select the right method for the patient population they investigating.

Description of the cohort

In all three studies, participants with an e-Boks account received an electronic survey via REDCap. Participants without an e-Boks account received an identical survey on paper by mail. Participants who had not responded to the initial e-Boks invitation were sent either a reminder letter by post with a link to complete the electronic survey (DANCAS, PReTT ) or a paper copy of the survey (NUTIR-HAB).

All e-Boks participants could request a paper questionnaire by phone or e-mail. All participants needed to have a Danish personal identification number, be resident in Denmark, ≥18 years of age, able to read and write in Danish, and not exempt from participation in scientific surveys. Specific eligibility criteria, identification of population and survey response rates were as follows:

DANCAS:

The Danish Out-of-Hospital Cardiac Arrest (DHRCA) registry and central personal register (CPR) was used to identify people who had suffered an OHCA (out-of-hospital cardiac arrest) from 1 January 2016 to 31st December 2019 and were alive at least 30-days post-cardiac arrest.

DANCAS survey data was collected from October 2020 to May 2021. Total eligible population was 2121 with 1266 respondents, giving an overall response rate of 60%.

PReTT:

The sample comprised of people with Parkinson's disease, who were identified through the National Patient Register, the Medicines Statistics Register and CPR. PReTT survey data was collected January to April 2020. Total eligible survey population was 11,770 with 7039 respondents, giving an overall response rate of 59.8%.

NUTRI-HAB:

The sample comprised of people with a diagnosis of oral, laryngeal, or pharyngeal cancer, and that had completed radiotherapy of curative intent between 1st of March 2014 and 28th of February 2018. They were identified through the Danish Head and Neck Cancer Group's (DAHANCA) national clinical quality database. Survey data collected March to June 2020. Total NUTRI-HAB survey study population was 1937, with 1190 respondents giving an overall response rate of 61.4%.

Data and biological material

Questionnaire data:

DANCAS

EQ-5D-5L, Hospital Anxiety and Depression Scale, Two Simple Questions, Modified Fatigue Impact Scale, 12-item World Health Organisation Disability Assessment Scale 2.0, physical activity questions, plus questions on unmet rehabilitation and information needs.

PReTT

Background information, time since diagnosis, EQ-5D-5L, PDQ-39, MDS-UPDRS, questions inspired by Hoehn & Yahrs scale, plus questions on rehabilitation received and unmet rehabilitation needs.

NUTRI-HAB

EQ-5D-5L, EORTC QLQ-C30, EORTC QLQ-H&N35, The Scored Patient-Generated Subjective Global Assessment Short Form, M. D. Anderson Dysphagia Inventory, plus questions on weight, weight history, dietary intake and health behavior, rehabilitation received, unmet rehabilitation needs and perceived support from and for relatives

Registry data:

Socioeconomic variables collected from registries will be age, gender and Danish Region, along with vital status, civil status, household status, education level, income, immigrant status, and comorbidities.

Publications associated with the project

Joshi VL, Tang LH, Borregaard B, et al. Long-term physical and psychological outcomes after out-of-hospital cardiac arrest-protocol for a national cross-sectional survey of survivors and their relatives (the DANCAS survey). BMJ Open 2021;11(4):e045668. doi: 10.1136/bmjopen-2020-045668 [published Online First: 2021/04/04]