5. RECRUITMENT RESULTS

Download Chapter 5 - RECRUITMENT RESULTS PDF 290kb

5.1 Korean War veterans
5.1.1 Reasons for non-participation in the Korean War veteran group
5.2 Population sample
5.2.1 Reasons for non-participation in the population sample
5.3 Investigation of possible participation bias
5.3.1 Comparison of participants and non-participants on demographic variables
5.3.2 Comparison of participants and study refusers on health information provided on the Voluntary Refusal Notification Form
5.4 Participation by proxy
5.5 Identification of the eligible comparison group from the population sample participants
5.6 Questionnaire data completeness and quality

Study recruitment commenced in March 2004 with the mailout of preliminary letters to all subjects. These were followed four weeks later by first-invitation packages, mailed in batches of approximately 2,000 per week (1,400 to Korean War veterans and 600 to population sample subjects). Non-responders received first-reminder letters three weeks after dispatch of their first invitation package, and second-reminder packages three weeks after dispatch of the first reminder letter. The last of the second-reminder packages were mailed in June 2004. Returned, completed questionnaires were accepted up until the end of August 2004, when recruitment closed.

5.1 Korean War veterans

The study commenced with 7,612 male Korean War veterans known or assumed to be alive and thought to be residing in Australia. During the recruitment period an additional 13 Korean War veterans, whose live status had been classified as ‘unknown’ in the Mortality Study [19], contacted the study team. They included two members of the population sample. All 13 veterans were added to the Korean War veteran study group, bringing the total to 7,625. Of this total, one Korean War veteran was identified as residing overseas during the data collection period, and 95 were reported deceased; these were removed from the group. Finally four subjects who had been included in the study were identified as not having deployed to the Korean War; these subjects were also excluded, bringing the final total to 7,525. The recruitment results for the 7,525 Korean War veterans are shown in Table 2.

Table 2 . Recruitment results for the Korean War veterans
 

Korean War veterans

Number

(%)

Final study group

7,525

-

Participants

6,122

(81.4)

Refusers

957

(12.7)

Non-responders

396

(5.3)

Not-contactable

50

(0.7)

6,122 Korean War veterans participated in the study by returning a completed questionnaire, representing an 81% participation rate in this study group. Whilst 957 veterans (13%) contacted the study team to indicate that they did not wish to participate, a further 396 veterans (5%) did not respond to their mailed invitations. Less than 1% of the posted questionnaires were returned to the study team as undeliverable (eg marked "not known at this address") and, if an accurate address could not be found, the subjects to whom these questionnaires were addressed were categorised as not-contactable.

[back to top]

5.1.1 Reasons for non-participation in the Korean War veteran group

Of the 957 Korean War veterans who declined participation in the study, 687 (76%) of these responded to a question on the Voluntary Refusal Notification Form asking them why they chose not to participate in the study. From a list of six options provided on the Form, to which refusers could select as many options as applied:

  • 12 (2%) Korean War veterans selected "you are too busy";
  • 204 (30%) selected "you are not well enough";
  • 208 (30%) selected "you don’t think the study applies to you";
  • 115 (17%) selected "the questionnaire is too long";
  • 150 (22%) selected "you are not interested in participating in a health study; and
  • 217 (32%) Korean War veterans selected "another reason" for non-participation and provided detail in relation to the latter option.

Reasons nominated by this latter group encompassed a number of different themes; including the assertion that the study was being conducted 50 years too late; that personal information was to remain private; that the items in the questionnaire were not of any use, or they were too intrusive, or that they favoured Army responders and not other Services; that the government could not be trusted; and that the individual’s service in Korea was too short to be of relevance, or not related to current health.

The reasons for non-participation by the 398 Korean War veteran non-responders is not known, however it is assumed that some may not have received a posted invitation at their correct address, and that others did not participate for similar reasons to those given above by the refusers.

[back to top]

5.2 Population sample

The study commenced with a population sample of 3,022 subjects. During the recruitment period one subject was identified as residing overseas, 43 were reported to be deceased, and these subjects were removed from the population sample. Two subjects were identified as Korean War veterans, and these two were removed from the population sample and added to the Korean War veteran group. An additional 9 subjects were removed as it was determined that they also did not meet eligibility criteria for inclusion in the population sample group; reasons for ineligibility included female sex and age less than 65 years. Recruitment results for the remaining 2,964 population sample subjects are shown in Table 3.

1,893 population sample subjects participated in the study by submitting a questionnaire, representing a recruitment rate of 64%. A further 21% of the population sample contacted the study team to decline participation, 13% did not respond to their mailed invitation, and 2% of the posted questionnaires were returned as undeliverable.

Table 3 . Recruitment results for the population sample
 

Population sample

Number

(%)

Final sample

2,964

-

Participants

1,893

(63.9)

Refusers

616

(20.8)

Non-responders

388

(13.1)

Not-contactable

67

(2.3)

[back to top]

5.2.1 Reasons for non-participation in the population sample

Of the 616 population sample subjects who declined participation in the study, 347 (62%) of these responded to a question on the Voluntary Refusal Notification Form asking them why they chose not to participate in the study. From a list of six options provided on the Form, to which refusers could select as many options as applied:

  • 30 (9%) population sample subjects selected "you are too busy";
  • 73 (21%) selected "you are not well enough";
  • 82 (24%) selected "you don’t think the study applies to you";
  • 54 (16%) selected "the questionnaire is too long";
  • 113 (33%) selected "you are not interested in participating in a health study; and
  • 78 (22%) selected "another reason".

Reasons provided by the latter group included language difficulties and lack of English, that the requested information was too private or personal, or that the subject’s ill wife needed to be cared for. A few population sample subjects responded that they did not participate because they were not Korean War veterans.

It is assumed that amongst the 387 population sample non-responders, the reasons for non-participation are similar to those given by the refusers above, and that some non-responders may not have received a posted invitation at their correct address.

[back to top]

5.3 Investigation of possible participation bias

Participation bias can occur if participants differ from non-participants on characteristics which are associated with the study dependent measures, such as health status. A complete examination of participation bias would require the collection of comprehensive health, demographics and deployment information for all non-participating Korean War veterans and comparison group subjects. Whilst complete data was not available for all non-participants, we were able to conduct several comparisons of data available for both study participants, and non-participants, to assess the level to which participants were representative of the study groups from which they were drawn.

[back to top]

5.3.1 Comparison of participants and non-participants on demographic variables

Within each study group, participants and non-participants were compared on several available demographic variables. For Korean War veterans, these variables included:

  • Age category based on date of birth in the DVA Korean War Nominal Roll.
  • State or Territory of residence based on postal address.
  • Service branch at the time of the Korean War.
  • Service rank at the time of the Korean War.

For the population sample, these variables included:

  • Age category drawn from the Electoral Roll.
  • State or Territory of residence based on postal address.

As shown in Table 4, participating Korean War veterans were very representative of the invited population based on the known demographic variables, with participation rates across most categories of each variable staying close to the overall population participation rate of approximately 81%. There were some exceptions.

Participation rates were very consistent across most age categories, with the exception of a decrease in participation noted amongst the oldest invited veterans aged 80 years or older.

Participation rates across the States contributing the largest numbers of veterans were similar to each other, at close to 81%, but there was some variability in the rates noted for the States or Territories contributing smaller numbers of veterans.

The participation rates amongst former Navy and Army personnel were close to 81%, whilst amongst former Air Force personnel the rate was a little higher at 87%.

Finally, Korean War veterans who had served in the war with enlisted ranks had a participation rate close to 81% whilst officers and non-commissioned officers had slightly higher participation rates.

Table 4 . Participation rates across age category, State or Territory of residence, Service branch and rank for Korean War veteran participants versus non-participants
 

Korean War veterans

Participants
(N=6,122)

Non-participants (N=1,403)

 


Total


n (%)


n (%)

Participation rate (% of total)

Age category

66-69

599

486 (7.9)

113 (8.1)

(81.1)

70-71

1,371

1,137 (18.6)

234 (16.7)

(82.9)

72-73

1,699

1,419 (23.2)

280 (20.0)

(83.5)

74-75

1,567

1,284 (21.0)

283 (20.2)

(81.9)

76-79

1,607

1,288 (21.0)

319 (22.7)

(80.1)

80-95

682

508 (8.3)

174 (12.4)

(74.5)

State or Territory

New South Wales

2,732

2,234 (36.5)

498 (35.5)

(81.8)

Queensland

1,745

1,402 (22.9)

343 (24.4)

(80.3)

Victoria

1,318

1,047 (17.1)

271 (19.3)

(79.4)

Western Australia

790

663 (10.8)

127 (9.1)

(83.9)

South Australia

425

373 (6.1)

52 (3.7)

(87.8)

Tasmania

241

191 (3.1)

50 (3.6)

(79.3)

Australian Capital Territory

199

176 (2.9)

23 (1.6)

(88.4)

Northern Territory

45

34 (0.6)

11 (0.8)

(75.6)

Service branch

Navy

2,800

2,310 (37.7)

490 (34.9)

(82.5)

Army

4,178

3,335 (54.5)

843 (60.1)

(79.8)

Air Force

547

477 (7.8)

70 (5.0)

(87.2)

Rank

Officer

524

444 (7.3)

80 (5.7)

(84.7)

Non-commissioned Officer

1,364

1,141 (18.7)

223 (15.9)

(83.7)

Enlisted rank

5,632

4,532 (74.1)

1,100 (78.4)

(80.5)

Table 5 shows that the population sample participants were also fairly representative of the larger, invited sample, however there was some increase in participation evident amongst the younger subjects and a decrease in participation evident amongst the oldest subjects.

Participation rates across most States and Territories varied little, with all except Tasmania staying close to the overall population participation rate of 64%.

Table 5 . Participation rates across age category and State or Territory of residence, for population sample participants versus non-participants
 

Population sample

Participants
(N=1,893)

Non-participants
(N=1,071)

 

Total

n (%)

n (%)

Participation rate (% of total)

Age category

66-69

500

399 (21.1)

201 (18.8)

(79.8)

70-71

414

287 (15.2)

127 (11.9)

(69.3)

72-73

501

320 (16.9)

181 (16.9)

(63.9)

74-75

464

299 (15.8)

165 (15.4)

(64.4)

76-79

497

303 (16.0)

194 (18.1)

(61.0)

80-99

488

285 (15.1)

203 (19.0)

(58.4)

State or Territory

New South Wales

1,012

627 (33.1)

385 (35.9)

(62.0)

Victoria

737

458 (24.2)

279 (26.1)

(62.1)

Queensland

550

357 (18.9)

193 (18.0)

(64.9)

South Australia

271

179 (9.5)

92 (8.6)

(66.1)

Western Australia

258

175 (9.2)

83 (7.7)

(67.8)

Tasmania

93

70 (3.7)

23 (2.1)

(75.3)

Australian Capital Territory

34

22 (1.2)

12 (1.1)

(64.7)

Northern Territory

8

5 (0.3)

3 (0.3)

(62.5)

[back to top]

5.3.2 Comparison of participants and study refusers on health information provided on the Voluntary Refusal Notification Form

The Voluntary Refusal Notification Form, which subjects could complete to advise the study team of their intention to decline participation in the study, is described in section 4.10.2 of this report and shown in Appendix J. The Form included the option for refusers to provide answers to two brief questions about their health. The questions were item 2 from the WHOQOL-Bref in regard to level of satisfaction with health, and also the Life Satisfaction scale.

Of the 957 Korean War veteran refusers, 588 (61%) provided answers to the health satisfaction item from the WHOQOL-Bref, and 591 (62%) provided answers to the Life Satisfaction scale on the Voluntary Refusal Notification Form. Their responses to these questions on the Form were compared with the responses of the Korean War veteran participants who answered the same questions within the participant questionnaire (n=6,095; 99.6%, who answered the WHOQOL-Bref health satisfaction question, and n=6,062; 99.0%, who answered Life Satisfaction). The age distribution of the Korean War veteran refusers, was found to be very similar to that of the larger group of Korean War non-participants (the latter group is shown in Table 4), of which the refusers are a part.

Of the 616 population sample refusers, 273 (44%) provided answers to the WHOQOL-Bref health satisfaction question, and 277 (45%) provided answers to the Life Satisfaction scale on the Voluntary Refusal Notification Form. Their responses to these questions on the Form were compared with the responses of the population sample participants who answered the same questions within the participant questionnaire (n=1,877; 99.2%, who answered WHOQOL-Bref health satisfaction, and n=1,886; 99.6%, who answered Life Satisfaction). The age distribution of the population sample refusers, was found to be very similar to that of the larger group of population sample non-participants (shown in Table 5), of which the refusers are a part.

The results for both comparisons are shown in Table 6. The results for the WHOQOL-Bref item are expressed as mean WHOQOL-Bref health satisfaction scores, with higher scores representing greater satisfaction with one’s health. The results for the Life Satisfaction scale are expressed as mean Percent Life Satisfaction scores, with higher scores representing greater life satisfaction.

Table 6 . Comparison of participants with study refusers who provided health information on the Voluntary Refusal Notification Form
 

Korean War veterans

Population sample

Participants

Refusers

Participants

Refusers

Mean

(SD)

Mean

(SD)

Mean

(SD)

Mean

(SD)

WHOQOL-Bref health satisfaction

3.0

(1.1)

2.7

(1.2)

3.7

(1.0)

3.5

(1.0)

Percent Life Satisfaction

55.8

(21.8)

49.8

(26.2)

68.4

(19.2)

63.1

(21.8)

The important comparison here is that between the participants and refusers within each study group, and not the comparison between groups. Within both the Korean War veteran group and the population sample, participants scored slightly higher on both the WHOQOL-Bref health satisfaction and Life Satisfaction measures than did the refusers. This finding implies that in both study groups, there was a similar pattern of the least well subjects being least likely to participate in the study.

The slight differences between participants and non-participants in age (shown above at section 5.3.1) and in average level of reported health and life satisfaction, is likely to result in the study slightly overestimating the true health of the Korean War veteran and comparison group populations; i.e. the observed results based on participant data are likely to be healthier, on average, than those which would have been observed if full participation in each group had occurred. However, because a similar pattern of non-participation amongst the oldest and most unwell occurred in both study groups, it is unlikely that these will notably effect the magnitude or direction of any differences in health outcomes between study groups which are observed in the study.

[back to top]

5.4 Participation by proxy

On page 2 of the questionnaire, 5,650 (92%) Korean War veteran respondents and 1,725 (91%) population sample respondents provided information about whether the questionnaire was being filled in by the invited participant (by self-report), or by a relative, carer or friend on behalf of the invited participant (by proxy). Some respondents indicated that the questionnaire was being completed both by self-report and by proxy. The percentages for these outcomes are shown in Table 7 for both the Korean War veteran group and the population sample.

Table 7 . Number and percentage of Korean War veteran and population sample participants who completed the questionnaire by self-report, by proxy, or both

Completed questionnaire by:

Korean War veterans
N=5,650

Population sample
N=1,725

 

n

(%)

n

(%)

Self report

5,129

(90.8)

1,485

(86.1)

Proxy

418

(7.4)

194

(11.2)

Both

103

(1.8)

46

(2.7)

Of the respondents shown in Table 7, 91% of Korean War veterans and 86% of population sample participants reported completing the questionnaire by self-report. Most of the respondents who reported that they completed the questionnaire by proxy, or by both self-report and proxy, then answered a question as to whether they transcribed answers provided by the participant, wrote their own answers on behalf of the participant, or both (transcribed and wrote own answers). These results are shown in Table 8.

Table 8 . Number and percentage of proxy, or both self-report and proxy, respondents, who transcribed answers provided by the participant, wrote their own answers on behalf of the participant, or both transcribed and wrote own answers
 

Korean War veterans
N=504

Population sample
N=233

 

n

(%)

n

(%)

Transcribed

430

(85.3)

184

(79.0)

Wrote own

70

(13.9)

44

(18.9)

Both

4

(0.8)

5

(2.1)

Only 70 Korean War veteran proxy respondents, and 44 population sample proxy respondents, reported that they wrote their own answers in the questionnaire on behalf of the invited participants. These represent only 1% and 2% respectively of all Korean War veteran and population sample participants.

An additional investigation was conducted to determine whether self-report respondents differed in age to those who responded by proxy. As shown in Table 9 for Korean War veterans, participants in the older age categories were less likely to participate by self-report and more likely to participate by proxy, compared with participants in the younger age categories. However even in the oldest age category, 85 years and older, a large majority of Korean War veteran participants (72%) responded by self-report. The results for the population sample followed a similar pattern, and are not tabulated.

Table 9 . Korean War veteran participation rates across age category: self-report versus proxy participants
 

All Korean War veteran participants

Korean War veterans participating by self-report
(N=5,129)

Korean War veterans participating by proxy
(N=418)

Age category

Total

n

(% participation rate)

n

(% participation rate)

66-69

358

322

(89.9)

13

(3.6)

70-73

2,411

2,048

(84.9)

129

(5.4)

74-79

2,773

2,313

(83.4)

210

(7.6)

80-84

435

348

(80.0)

45

(10.3)

85+

105

76

(72.4)

17

(16.2)

Based on the above assessment of proxy participation it was determined that it was unlikely that misclassification of participant health, resulting from proxies erroneously estimating questionnaire responses, would notably impact upon the study results. This was because only a small proportion of questionnaire responses in both groups were provided by proxy, and because most proxy respondents transcribed answers provided by the invited participant, rather than estimating their own answers on behalf of the participant.

[back to top]

5.5 Identification of the eligible comparison group from the population sample participants

The responses provided by the 1,893 population sample participants to questions 2, 2a and 2b in their participant questionnaire, were reviewed to determine which participants were eligible for inclusion in the study’s final comparison group. These questions pertained to country of birth, year of first settlement in Australia and citizenship status. Eligibility criteria for membership of the study comparison group is detailed in section 4.8.2 and included that participants be:

  • Australian born; or
  • if not Australian born, then first settled in, or a citizen of, Australia by 1955 or earlier.

Table 10 shows that 1,510 (80%) population sample subjects met criteria for comparison group membership. It is the results for these 1,510 comparison group participants which will be compared with the results of the Korean War veteran participants, and which will be reported throughout Chapter 6 Participant results: Korean War veterans versus comparison group.

Table 10 also shows that 357 (19%) of the population sample subjects were assessed as not eligible for the comparison group, being neither Australian born nor first settled in, nor citizens of, Australia by 1955 or earlier. In the case of a further 26 (1%) population sample subjects, insufficient information was provided at questions 2, 2a and 2b to determine eligibility.

Table 10 . Population sample participants assessed as eligible or not eligible for comparison group inclusion
 

Population sample (N=1,893)

n

(%)

Total eligible for inclusion in comparison group

1,510

(79.8)

Total not eligible for inclusion in comparison group

357

(18.9)

Eligibility undetermined

26

(1.4)

[back to top]

5.6 Questionnaire data completeness and quality

Participants in both study groups provided very complete questionnaire data with very few missing responses in most sections. Between 95% and 99% of all participants provided sufficiently complete data to be scored on measures of smoking and alcohol consumption, and on the HAD, Life Satisfaction and WHOQOL-Bref questionnaires. Approximately 91% of all participants could be scored on the PCL. Data provided in relation to self-reported medical conditions was less complete, with up to 17% of participants not providing answers in relation to some medical conditions.

Checking of the quality of Datatime Pty Ltd’s electronically entered data for 50 Korean War veteran questionnaires revealed transcription errors in 2.4% of all entered text fields and in 0.3% of all scanned tick-box fields, whilst checking of electronically entered data for 50 population sample questionnaires revealed errors in 0.2% of entered text fields and 0.04% of scanned tick-box fields. Overall, checking of the electronic data entry revealed transcription errors in only 0.4% of all fields.

Statistical checks to detect invalid, inconsistent or outlying results included range checks for participant age at time of questionnaire completion, or at time of war deployment, and checks for questionnaire responses within the ranges allowed by individual instruments. It was observed that invalid, out of range data primarily resulted from text-field data entry error, and that these could often be corrected after referring back to the original questionnaires.