• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of injprevInjury PreventionCurrent TOCInstructions for authors
Inj Prev. Apr 2007; 13(2): 88–92.
PMCID: PMC2610597

Challenges of recruiting farm injury study participants through hospital emergency departments

Abstract

Background

Hospital emergency departments are common recruitment sites for injury studies. Yet recruitment method details, capture rates and response fractions are not consistently reported. As privacy legislation increasingly impinges on research activity, these parameters become even more important. The authors describe their experience with recruitment via emergency departments and outline subsequent adjustments to the recruitment approach.

Methods

The FIRM study was an Australian case‐control study of serious farm work‐related injury. Cases were identified prospectively by hospital staff on presentation to emergency departments. Consistent with the Victorian Health Records Act, potential cases were initially approached by hospital staff, and full recruitment was subsequently undertaken by study staff. Manual hospital record audits were conducted at five recruitment sites to determine the proportion of eligible cases approached.

Results

Among 660 medical records audited, 19 eligible cases were confirmed, 9 of whom were approached by hospital staff (47%, 95% CI 25 to 70%). In response, an additional process was established to capture missed cases, who were sent a letter from the hospital providing the opportunity to opt out of telephone contact by study staff. Early indications were that 34% (41/122) of missed cases actively declined to be contacted. Among those who were contacted and eligible, 84% (21/25) agree to study participation.

Conclusions

Recruitment of injury research participants via hospital emergency departments is challenging, particularly where authorities require an intermediary to make the initial contact. Removal of some constraints imposed by privacy legislation would considerably simplify recruitment and enhance scientific rigour in conducting epidemiological research.

Hospital emergency departments are common recruitment sites for injury research. Failure to identify and approach eligible patients could threaten successful recruitment, particularly when combined with low response rates. Few studies report these parameters, or provide detailed descriptions of recruitment processes.1

Increasing concerns about privacy and ethics around the world are leading to legislative changes, some of which appear to be adversely impacting health research. Recruitment methods that rely on accessing health and/or personal information held by a third party, such as hospitals, are particularly vulnerable. Australia is a case in point. Personal and health information that identifies an individual cannot be used or released without consent, with a small number of exceptions.2 Researchers have consequently lost the ability to contact participants directly from data registries, potentially compromising scientific rigour for reasons such as biased sampling, inability to describe the pool of all eligible participants, and misclassification of exposure or outcome.2,3 These concerns are shared across international boundaries with evidence emerging from the US and the UK that recruitment methods required by privacy legislation adversely impact on response rates and introduce significant response bias, in addition to increasing the cost and complexity of epidemiological research.4,5

In selecting hospital emergency departments as the site for case recruitment in our case control study of farm injury, we were aware of the need to maximise the identification of, and approach to, eligible patients. Demand for emergency department services is often high,6 creating a busy work environment. Farm injury accounts for a low proportion of emergency department presentations even in rural hospitals,7,8 requiring a good alerting system when an eligible patient presents. In addition, to achieve our sample size, recruitment involved a number of geographically disparate hospitals, each with its own human research ethics committee and different emergency department staffing structures. The aims of this paper are to (1) describe the initial method for case recruitment, (2) describe the outcome of this method, including an audit of the recruitment approach, and (3) outline an adjusted recruitment approach and its result. The initial case recruitment process and the audit procedure are outlined in the Methods section, while the related results and the adjusted recruitment approach are included in the Results section.

Methods

The Farm Injury Risk among Men study

The Farm Injury Risk among Men (FIRM) study was a case‐control study of postulated host and farm risk factors for serious farm work‐related injury.9 The study base was adult males working on farms in the catchment areas (defined by zipcode) of 14 Victorian regional hospitals. Eligibility criteria for cases included: over 16 years of age; injured while working on a study region farm; injury scoring 2 or higher on a measure of physical damage, the Abbreviated Injury Scale (AIS).10

Initial method of identifying and recruiting cases

Non‐fatally injured cases were recruited prospectively through emergency departments of 14 regional hospitals, and admitting units of five metropolitan hospitals to which these regional hospitals referred the most serious cases to access major trauma centers and specialist services such as microsurgery and ophthalmology. The 19 study hospitals treat 74% of farm work‐related injuries occurring in the study region that are admitted to hospitals in Victoria (unpublished data from the Victorian Admitted Episodes Database).

The regional study hospitals included all major rural hospitals across Victoria (87 649 square miles), and several key smaller hospitals serving farming communities. In 2004/5 presentations averaged 21 690 per emergency department (range 4471–38 545; personal communication, Peter Darby, Victorian Department of Human Services, 2005). Of presentations to these hospitals 0.3% were potentially eligible for FIRM. We aimed to approach all eligible patients from 2003–5.

At study commencement, all except four of these hospitals had an electronic injury surveillance system and contributed to the Victorian Emergency Minimum Dataset (VEMD) under an agreement with the Victorian Department of Human Services. The VEMD records details of injuries treated at hospital emergency departments according to the National Data Standards for Injury Surveillance.11 Each hospital is responsible for training staff to collect VEMD data, and receives regular feedback on data completeness and quality. Details on the circumstances surrounding each presentation, including activity at the time (eg, work) and location (eg, farm), is generally sought. Collection of these variables is not mandatory, and patient treatment and completion of the electronic record can proceed without these data having been entered onto the system.

The Health Records Act 2001 (Victoria)12 defines a number of Health Privacy Principles, one of which relates to the disclosure of health information for a secondary purpose, such as research. The initial guidelines published by the Health Services Commissioner13 required that screening of patients for research purposes should generally be undertaken by hospital staff, although this is not explicitly stated in the Act. Consequently, in the FIRM protocol, potentially eligible patients (male, at least 16 years of age, injured while working on a farm) were approached by hospital staff seeking approval to give patient contact details to Monash University for the purposes of contacting them regarding the study. This could occur either at the point of registration, or during their time in the emergency department. Once consent to contact was given, the remainder of the screening and recruitment process was conducted by FIRM study staff. Injury severity was determined by study staff, and eligible patients were invited to participate (fig 11).

figure ip13110.f1
Figure 1 Recruitment of non‐fatally injured cases, February 2003–February 2004, Farm Injury Risk among Men study, Victoria, Australia.

The FIRM recruitment process required recognition of a relatively rare event by busy emergency department staff, so measures were implemented to maintain the study profile. FIRM study staff delivered ongoing education and training sessions regarding eligibility criteria at all study hospitals, and strategically located posters reminded staff of the eligibility criteria and the consent to contact process. Regular updates were provided for staff notice boards. Posters were placed in the waiting and treatment areas of all participating hospitals.

The study had approval from the Monash University Standing Committee on Ethics in Research involving Humans, and the 17 human ethics research committees responsible for the 19 study hospitals. The average ethics approval period was 77 days, ranging from 23–245 days.

Case recruitment audit

A hospital record audit was conducted early in the anticipated three‐year recruitment period at the first five hospitals to start recruitment. The aim was to gauge the proportion of eligible patients presenting to these emergency department who were approached for participation. These five hospitals treated slightly more patients in their emergency departments than all regional study hospitals (26 717 compared with 21 690 in 2004/5). Of all presentations to the audited hospitals, 1.2% were eligible for FIRM compared with 0.3% overall for the regional hospitals.

Cases recruited at the audited hospitals were similar with respect to injury severity, farming experience, and education level compared with those recruited at all regional hospitals. They were slightly younger (29% over 55 years of age compared with 36%).

Appropriate presentations for audit would be patients meeting all five eligibility criteria (male, 16 years+, injured on a farm, during farm work‐related activity, AIS 2+). The first four criteria are variables of the injury surveillance system, and define potentially eligible patients. Theoretically, medical records for auditing were those for patients identified from the injury surveillance system by these four key variables who also had an AIS 2+. However, previous auditing of the injury surveillance system indicated that location and activity at the time of injury may be miscoded. For example, an injury occurring on a farm may be coded as occurring at home, or unpaid work may be coded as recreation (personal communication, Karen Ashby, Monash University, 2005), or unspecified codes may be selected.7,14 Among all injury emergency department presentations in Victoria, 15% have an unspecified code for location of injury.14

We elected to audit a wider set of records than just those meeting our eligibility criteria. A sample was created of male emergency department presentations aged at least 16 years for the period February to April 2003, presenting with an unintentional injury sustained in various locations (farm, home, industrial or construction areas, other specified and unspecified locations), and injured during various activities (working for income, other work, leisure, other specified and unspecified activities). This sample contained 2335 patients, too many for manual auditing within resources. To select a subsample which maximized the number of eligible presentations, both contacted and non‐contacted, we selected from the 2335 patients (a) all presentations coded as farm location (n = 149), (b) all presentations occurring on the same dates as patients who had consented to contact (n = 138), (c) a random selection from the remainder (n = 373). The patients in this random selection were similar to the original 2335 patients with respect to age, injury cause, and admission status.

Manual checking of these 660 patient medical records identified patients meeting the case eligibility criteria (including the injury severity criterion). For these patients, relevant details (de‐identified in accordance with privacy requirements) were extracted and checked against study consent to contact forms to determine the proportion that had been approached for study participation.

Results

Outcome of initial recruitment method

Recruitment began at the first hospital in January 2003, with all hospitals recruiting by February 2004. Hospital staff approached 281 potentially eligible patients, of whom 74 were contactable and eligible (fig 11).). If the proportion meeting the injury severity definition was similar between those who were contacted, and those who declined contact or were not contactable, then an estimated 129 eligible cases were approached by hospital staff for consent to contact. Of these, 64 (50%) consented and were interviewed. In the regional hospitals, an estimated 116 eligible cases were approached and 55 (47%) consented and participated.

Case recruitment audit results

Among 660 audited records, 19 patients met all eligibility criteria, 9 (47%, 95% CI 25 to 70%) of whom had received a consent‐to‐contact form from emergency department staff (fig 22).). The remaining 10 patients were missed by the consent‐to‐contact process.

figure ip13110.f2
Figure 2 Audit of recruitment approach, February–April 2003, Farm Injury Risk among Men study, Victoria, Australia.

Adjustments to the initial recruitment method

The audit results suggested that a high proportion of eligible patients were not being approached in the emergency department, potentially threatening the scientific merit of the FIRM study via detection and response bias. In addition, the first 12 months of FIRM had a modest response rate that, if continued, would have extended the recruitment period well beyond 2005, incurring unbudgeted costs, and delaying dissemination of the results.

Therefore, following the audit, a modified recruitment process was developed. Hospital staff generated regular (eg, monthly) lists of potentially eligible cases using the injury surveillance system to identify males, 16 years and over, presenting with an unintentional injury sustained on a farm during work‐related activities. Patients already approached would be removed from the lists, and the remainder sent a letter from the hospital providing the opportunity to opt out of telephone contact by study staff. The Office of the Health Services Commissioner was consulted to ensure this process was consistent with the Health Records Act. A methodological drawback of this approach was that it increased the time between injury occurrence and interview.

Requests for this amendment to recruitment were made to the human ethics committees of Monash University and the eight regional recruiting sites that had an electronic surveillance system and were willing to implement the process. All requests were approved, except one where the human ethics committee approved an opt‐in process whereby the patient was invited to return a form directly to Monash University by mail if interested in participating. This preference for an opt‐in rather than opt‐out process seems to be reflective of trends observed in other countries.15

Early indications following the implementation of the new recruitment process were that 34% (41/122) of initially missed patients actively declined to be contacted when sent the letter. Among the remainder who were contactable, 65% (47/72) were ineligible (ie, AIS <2). Of the eligible 25 patients, 21 (84%) agreed to participate. These response results are similar to patients initially approached in the emergency department (fig 11).). The new process will significantly increase the recruitment rate at the seven sites where it can be implemented, because the number of potential participants approached will be increased. These sites recruited 62% of cases before implementation of the new recruitment process.

Discussion

The initial process for recruiting injured farmers for our case‐control study through hospital emergency departments may have missed nearly 50% of eligible cases, as a result of potentially eligible patients not being approached to participate. This is similar to the 44% of missed eligible cases reported by the only other published study using a similar recruitment method, and undertaking an audit process to ascertain capture.16 Our reminder systems (posters, regular verbal and written updates, nominated contact person) were insufficient to overcome the requirement of busy emergency department staff to recognize the relatively unusual presentation of an injured male farmer, and approach them for consent for contact by study staff. A direct personal approach by study staff to eligible patients using contact information provided by hospital data registries, which would have overcome these issues, was not consistent with the requirements of privacy legislation and its interpretation.

There are some limitations to our approach. The generalizability of audit results to all regional study hospitals is difficult to assess. Audited hospitals tended to treat more patients among whom the incidence of farm work‐related injury was higher. There were also small demographic differences. There were no data on which to judge the impact of these differences on case identification and approach. Further, we were unable to determine if the high proportion of potential cases who were missed resulted in any systematic bias in case recruitment. While our study is illustrative of the potential impact of similar privacy legislation in other countries it ultimately represents the Australian context only.

The implications for our study, and others using similar recruitment methods, are twofold. First, scientific rigour will be compromised via detection and response bias if the approached and non‐approached cases, and the participating and non‐participating cases differ on key characteristics of interest. Second, the time required to achieve sample size targets is greatly increased, with implications for project and budgetary management, and delays in the results being available to guide prevention programs. A modified recruitment process, as now being implemented with promise in the key FIRM regional hospitals, will increase the proportion of eligible patients who are approached to participate, thereby moderating the potential impact of these biases.

Removal of the constraints imposed by the privacy legislation would considerably simplify recruitment and enhance scientific rigour for future studies. Access to information, including identifying information, to allow adequate description of the entire pool of eligible participants and comparisons of approached and non‐approached, and participants and non‐participants, must be facilitated in Australia and elsewhere—by legislative amendment if necessary. Similarly, the opportunity for direct contact between researcher and potential participant for the purposes of recruitment should be reinstated. In our previous studies where recruitment of injured cases was by direct contact from a hospital register, response rates of non‐farmer groups ranged from 65–91%.17,18,19,20,21 In a study of injured farmers recruited by direct contact from a hospital register, the response rate was 98%.22 Our collective experience before the enactment of privacy legislation was that very few people believed contact by researchers for recruitment purposes was an invasion of privacy. The patient complaint officers of four of the busiest regional hospitals in our study reported virtually no complaints from patients as a result of being approached for participation in research. Similarly, a small proportion (1–2%) of complaints received by the Victorian Health Services Commissioner each year relate to health information being used by researchers to contact people about participation in research (Michael McDonald, personal communication, 2006). A recent study in the UK in which direct contact was allowed between researchers and potential participants (n = 12 500), found that only a small number queried the researcher's right to approach them.23

Provision of an opt‐out clause covering contact regarding research studies to all patients entering the hospital system could be explored. Reform of human ethics committee systems to streamline multicenter research, provision of clear definitions of terms such as “practicable” and “public good”, and intensive training of ethics committee members in interpretation of privacy legislation is also required. Investment in education programs to raise awareness in the community about the importance of giving due consideration to participation in accredited research also appears to be justified.

Key points

  • Details of methods for recruitment of injury study participants in hospital emergency departments are not commonly reported.
  • Prospective case identification by hospital emergency staff for this case‐control study resulted in about 50% of eligible cases not being approached.
  • A direct personal approach by study staff using contact information held by the hospital was not possible due to privacy legislation, an increasingly common circumstance in many countries.
  • A modified recruitment process involving letters subsequently sent to the missed eligible cases by the hospitals increased the recruitment rate.
  • Removal of some constraints imposed by privacy legislation would simplify recruitment and enhance rigour in epidemiological research.

In countries where similar privacy legislation continues to exist, injury researchers are well advised to consider alternative methods of case recruitment through emergency departments, and to ensure that adequate resources for case recruitment are included in project budgets. Dedicated recruiters in hospital emergency departments seem to improve the proportions of eligible patients approached regarding research studies. The availability of dedicated recruiters for 16 h per day for a minimum of five days per week has been associated with between 76 and 95% of eligible patients being approached regarding participation.24,25 The resource implications of this approach are obvious, particularly if multiple sites are involved. Financial incentives for emergency department staff could be considered, but would have similar resource implications. In the local context, monthly opt‐out letters from the hospital to eligible patients is a relatively low cost option, provided that eligibility can be determined from hospital databases and a time period of at least two months between the injury event and research data collection is methodologically acceptable.

The establishment of automated systems in large research active hospitals may also be considered. Such systems require computer programs which recognise eligibility criteria when entered into hospital databases and dispatch an electronic message (page, text message or email) to research staff, and have achieved notification of 84% of eligible patients.16

Acknowledgements

We are grateful to the other 13 members of the study team, the 19 participating hospitals, the Victorian Injury Surveillance and Applied Research Program, the Victorian Farmers Federation, and the study participants for their support and assistance. Wendy Watson and Carolyn Staines provided valuable comment on the draft manuscript.

Abbreviations

AIS - Abbreviated Injury Scale

FIRM - Farm Injury Risk among Men study

VEMD - Victorian Emergency Minimum Dataset

Footnotes

Source of funding: Project grant from Australian National Health and Medical Research Council. LD supported by Senior Research Fellowship from the Council.

Competing interests: None.

References

1. Olson S H, Voigt L F, Begg C B. et al Reporting participation in case‐control studies. Epidemiology 2002. 13123–126.126 [PubMed]
2. Whiteman D, Clutton C, Hill D. Privacy and epidemiology: report and perspective on the NHMRC's submission to the Review of Privacy Act. Australas Epidemiologist 2005. 124–8.8
3. O'Grady K F, Nolan T M. Privacy: bad for your health? Med J Australia 2004. 180307–308.308 [PubMed]
4. Angus V C, Entwistle V A, Emslie M J. et al The requirements for prior consent to participate on survey response rates: a population based survey in Grampian. BMC Health Serv Res 2003. 321–31.31 [PMC free article] [PubMed]
5. Armstrong D, Kilne‐Rogers E, Jani S M. et al Potential impact of HIPAA privacy rule on data collection in a registry of patients with acute coronary syndrome. Arch Intern Med 2005. 1651125–1129.1129 [PubMed]
6. Victorian Government Health Information Hospital Services Report. Department of Human Services, 2005. Available at http://www.health.vic.gov.au/hsr/index.htm (accessed October 2005)
7. Day L, Ashby K, Stathakis V. Unintentional farm injury. Hazard Edition No 33. Melbourne, Australia: Monash University Accident Research Centre, 1997
8. Harper C. Epidemiology of injuries in central Queensland: rural injury surveillance. Rockhampton, Australia: Central Public Health Unit, Queensland Health, 1997
9. Day L, Langley J, Voaklander D. et al Minimising bias in a case control study of farm injury. J Agric Saf Health 2005. 11175–84 http://monash.edu.au/muarc/projects/rural.html.84 http://monash.edu.au/muarc/projects/rural.html [PubMed]
10. Association for the Advancement of Automotive Medicine The Abbreviated Injury Scale. 1990 Revision. Barrington, IL USA: AAAM, 2001
11. Marson R, Taylor Mc D, Ashby K. et al Victorian Emergency Minimum Dataset: factors that impact upon the data quality. Emerg Med Australas 2005. 17104–112.112 [PubMed]
12. Office of the Health Services Commissioner (Victoria) Health Records Act 2001 (Vic). Statutory guidelines on research. Melbourne, Australia 2002
13. Office of the Health Services Commissioner (Victoria) Information Sheet No 7. “Screening” the health information of individuals to obtain consent to participate in research. Melbourne, Australia 2003
14. Clapperton A, Ashby K, Cassell E. Injury profile, Victoria 2001. Hazard Edition No 54. Melbourne, Australia: Monash University Accident Research Centre, 2003
15. Hewison J, Haines A. Overcoming barriers to recruitment in health research. BMJ 2006. 333300–302.302 [PMC free article] [PubMed]
16. Weiner D L, Butte A J, Hibberd P L. et al Computerized recruiting for clinical trials in real time. Ann Emerg Med 2003. 41242–246.246 [PubMed]
17. Day L, Ozanne‐Smith J, Parsons B. et al Eucalyptus oil poisoning among young children. Aust N Z J Pub Health 1997. 21297–302.302 [PubMed]
18. Voaklander D C, Kelly K D, Sukrani N. et al Sledding injuries presenting to the emergency department in a northern city. Academic Emerg Med 2001. 8629–635.635 [PubMed]
19. Kelly K D, Voaklander D C, Johnston W C. et al Equity in waiting times for major point arthroplasty. Can J Surg 2002. 45269–276.276 [PMC free article] [PubMed]
20. Sherker S, Ozanne‐Smith J, Rechnitzer G. et al Out on a limb: risk factors for arm fracture in playground equipment falls. Inj Prev 2005. 11120–124.124 [PMC free article] [PubMed]
21. Watson W. Of life and limb: measuring the burden of non‐fatal injury. PhD thesis. Melbourne, Australia: University of Monash, 2005
22. Layde P M, Nordstrom D L, Stueland D. et al Machine‐related occupational injuries in farm residents. Ann Epidemiol 1995. 5419–426.426 [PubMed]
23. Inskip H, Godfrey K, Robinson S. et al Cohort profile: The Southhampton women's survey. Int J Epidemiol 2005. 1–7.7 [PubMed]
24. Smith S R, Jaffe D M, Petty M. et al Recruitment into a long‐term pediatric asthma study during emergency department visits. J Asthma 2004. 41477–484.484 [PubMed]
25. Hollander J E, Sparano D M, Karounos M. et al Studies in emergency department data collection: shared versus split responsibility for patient enrolment. Acad Emerg Med 2004. 11200–203.203 [PubMed]

Articles from Injury Prevention are provided here courtesy of BMJ Group
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...