Engaging older Australians in sexual health research: SHAPE2 survey recruitment and sample

SHAPE2 study design

The SHAPE2 online survey employed a cross-sectional design, which is reported here guided by the CHERRIES checklist criteria for internet survey reporting.³⁵ The survey was designed using Qualtrics software and included 63 questions, each on an individual screen, organised into nine blocks (see Supplementary material, Appendix A for survey questions). We used adaptive questioning, whereby some questions were displayed conditionally depending on previous responses, ensuring all survey items displayed to participants were relevant. The survey was tested for usability and technical functionality by colleagues of the researchers and was then pilot tested with approximately 10 older adults, before final refinements. We initially designed this study as an online survey because of logistical difficulties with administering hard copy surveys during the coronavirus disease 2019 (COVID-19) pandemic and due to budget constraints, however, later in the data collection process a small number of paper surveys were posted out to those who contacted the researchers requesting hard copies. The online survey took participants a median of 17 min to complete.

The survey included questions on sexual health information-seeking as well as on demographics, relationships, identity, and behaviours. The questions on sexual health information-seeking included whether participants had sought sexual health information since turning 60 years old, what sources they had accessed, what sexual health topics they had inquired about, whether they had found answers to their questions, where they would look for sexual health information in future, what topics they would like to know more about and what makes sexual health resources trustworthy and appealing (Supplementary material, Appendix A).

Population and sampling

Like both SA&M and the original SHAPE Project, an age of 60 years or over and Australian residency were the only inclusion criteria, thus allowing for greater comparison with these earlier studies. Our aim was to recruit a sample size of 500 participants over 2 months to obtain confidence intervals of approximately ±4% around a proportion of 50%. However, we adjusted our target upwards and extended our recruitment timeline after strong initial response numbers.

We sought to recruit a sample as representative of older Australians as possible within the limitations of a convenience sample. We monitored basic demographic characteristics of participants as survey responses were submitted (age, location, gender, sexual orientation) and adjusted our recruitment strategy where necessary to recruit as diverse and representative a sample as was feasible. This study was undertaken on a modest budget of AUD400, which was used for paid Facebook advertising.

Study ethics

Ethical approval was granted by the University of Melbourne Human Research Ethics Committee (HREC ID 2057393). Prior to survey commencement, prospective participants were provided with a plain language statement, which included information about the study, expected time commitment, contact details for the researchers, information about data storage and security, and support services in case the survey brought up anything troubling for them. Before commencing, participants provided consent and proceeded to the survey questions; those that declined were exited from the survey. The survey was anonymous, and participants were not asked any identifying information.

Respondents could review and change previous responses using a ‘back’ button, skip questions they preferred not to answer and could exit the survey at any time. As the survey was anonymous; however, responses that were already entered could not be withdrawn. During data collection we checked responses to free-text questions periodically to identify any potential signs of distress (e.g. reference to past sexual assault). A small number of comments were identified as potentially indicating distress and were brought to the attention of the ethics committee. However, follow up with these respondents was not possible due to the survey’s anonymous design.

Recruitment

The survey opened on the 31 March 2021 with an initial soft release where it was promoted within researchers’ networks and on Twitter. Recruitment began in earnest on 7 April 2021 and was conducted in three staggered phases. Phase 1 (7 April–20 May 2021) involved emailing organisations that serve or attract older adults; Phase 2 (8 May–7 June 2021) used paid Facebook advertising targeting the population of interest; and Phase 3 (8 June–31 September) was the passive recruitment phase where the survey remained open, but active recruitment ceased. No incentives were given for participation.

During Phase 1, emails were sent in batches to approximately 2500 organisations, community groups (e.g. Council on the Ageing (COTA) Victoria, Probus, University of the Third Age, Men’s Sheds) and social groups via government councils. These contacts were found via internet searches. The email explained the study and the recruitment flyer was attached (Fig. 1). Recipients were encouraged to share the link and flyer with their members or communities as they saw fit (e.g. via newsletters, social media, or noticeboards). While we contacted online dating sites in Phase 1, none of them promoted the SHAPE2 survey.

Fig. 1.

SHAPE2 survey recruitment flyer.

For Phase 2, we launched a paid Facebook advertisement that was displayed only to older adults within Australia. After a week, we added a second advertisement to be displayed only to older women in Australia in an effort to recruit more women. We ran both advertisements concurrently for 2 weeks. For the final week, we stopped the second advertisement and continued only with the general advertisement, for which we changed the geographical settings so it was shown only in three States/Territories where we sought to increase participation. After 4 weeks, Facebook advertising ended as our AUD400 budget was reached.

After active recruitment ceased, we continued to record a small number of survey responses every few days and prospective participants continued to make contact by email and phone. Because of this, we left the survey open and allowed responses to trickled in, forming Phase 3 of recruitment. After 3 months of this passive recruitment, the survey was closed on 30 September 2021 after it had been open for a total of 6 months. This is the same timeframe for data collection as SA&M, providing opportunity for comparison.

Data presentation

We use descriptive statistics to present the characteristics of the SHAPE2 sample, comparing this with population data and the SA&M study sample, and we outline the differences between SHAPE2 participants based on how they were recruited into our study.

Sample recruited

We successfully recruited a sample with approximately equal numbers of male and female participants, that included ages 60–92 years (median age 69 years), and with respondents from throughout Australia. Participant characteristics are in Table 2 and comparisons made with Australian Bureau of Statistics (ABS) population data, and the SA&M sample from 2015. SHAPE2 participants are further broken down by whether they were recruited via Facebook or by other means.

Postcodes were compared to ABS data to determine remoteness³⁶ and socio-economic deciles.³⁷ Close to two thirds of survey participants were located in major cities (63.8%), approximately a third in regional locations (35.0%), and a small number in remote locations (1.2%) (Table 3). Postcodes from all socio-economic deciles were represented in the sample but skewed towards more advantaged geographical areas (Table 3).

Two thirds of respondents were retired (65.6%), one quarter were working either full-time, part-time or casually (27.1%), and the remainder designated other employment statuses (Table 3). Participants were able to state their occupation in a free-text question and gave a wide range of current or past jobs including cafe owner, dentist, taxi driver, teacher, public servant, mail sorter, corrections officer, farmer, flight attendant, nurse, and minister of religion among others.

Of those who answered the sexual orientation question (73.8% of participants), the majority were heterosexual (85.2%), and 11.2% either homosexual or bisexual (Table 3). Of those who answered the relationship status question (73.9% of participants), three out of five (61.8%) were in a cohabiting relationship with their spouse or partner, 19.8% were single and either dating or not-dating, while 8.5% were in a committed relationship but living separately, sometimes termed a LAT relationship (living apart together)³⁸ (Table 3). It was not feasible to compare with ABS or SA&M as these data were categorised differently across datasets.

Sample demographics

Social media vs other recruitment methods

Many older Australians are online, particularly the ‘younger-old’ aged in their 60s and 70s⁴⁰ and many are using social media, with Facebook the most popular platform for older age groups.⁴⁶ Just over a third of our sample found out about the survey via Facebook, contributing to the evidence that social media is effective for recruiting older people for sexual health surveys,^30,45 much as it has been shown to be effective for recruiting young people for sexual health research^47,48 and for recruiting diverse and hard to reach populations for health research broadly.⁴⁹

A greater proportion of our sample, however, came via newsletters/organisations or clubs (contacted by email during Phase 1). This is in contrast to SA&M, which recruited around half their sample from Facebook and a smaller proportion from newsletters/organisations or clubs. These differences in recruitment methods may contribute to why our sample was a better fit to ABS population data. While our email campaign was effective, it was not highly efficient. During Phase 1, recruitment we sent out approximately 2500 emails. Given that only 541 people indicated they found out about the survey via a newsletter or group, this shows that this recruitment method is labour-intensive and low yield, with most emails generating no response. Additionally, a potential limitation of the SHAPE2 sample is that it may over-represent the types of people who are involved in community groups and clubs.

By the end of recruitment Phases 1 and 2, we had recruited 1417 participants, representing 96% of our final sample. The low number recruited during Phase 3 of passive recruitment shows that few people accessed the survey through onwards-sharing after the active recruitment phase, demonstrating minimal benefit from leaving the survey open.

Strengths and limitations

A strength of SHAPE2 recruitment was the low budget used to recruit a sizeable sample. Our experience shows that a mixture of email promotion combined with Facebook advertising is not only feasible, but also economical for recruitment. While the reach of paid Facebook advertising is dependent on the amount of money spent, even our modest budget yielded good results. The SA&M researchers also found Facebook advertising to be ‘a cost-effective strategy’.³⁰

While our recruitment strategies were effective, there are some limitations to our sample and its generalisability, as have been observed in other internet surveys of older adults.⁵¹ Older Australians are using the internet more than ever however not all are online, particularly those aged 85 years and over.⁴⁰ As with other online-only surveys of older adults, people who are not online would not have been able to participate.³⁹ One prospective participant who saw the flyer posted on a community noticeboard phoned to request paper copies for themself and others who did not have internet access. We posted out these hard copies, however, it would not have been financially feasible to include hard copy distribution as a core component of the recruitment strategy. There may also have been others without internet access who were interested, but who did not contact us. It is unclear, however, how effective it would have been to include hard copy distribution as part of our recruitment strategy. SA&M distributed hard copies, but few were returned, with the vast majority completing the survey online.³⁰ This suggests that paper surveys are not preferred by many older adults, and those who are harder to reach would require a more resource-intensive approach to access them beyond the mere provision of a hard copy format. Including reply-paid envelopes was likely an important factor in the majority (96.6%) of 1876 survey responses of Australian retirement village residents being returned in paper format (only 3.4% online) in a 2015 study.⁵²

A limitation of the SHAPE2 study is that it was only available in English; thus, people who do not read and write English confidently would have been unable to participate. Additionally, we did not include a question about ethnicity or country of birth in the demographic questions for our survey and as a result we do not know how representative the sample was in this respect.