Combining conditional and unconditional recruitment incentives could facilitate telephone tracing in surveys of postpartum women

Journal of Clinical Epidemiology 59 (2006) 732–738

Combining conditional and unconditional recruitment incentives could facilitate telephone tracing in surveys of postpartum women Hind Beydouna, Audrey F. Saftlasa,*, Kari Harlanda, Elizabeth Tricheb a

Department of Epidemiology, College of Public Health, University of Iowa, GH C21-F, 200 Hawkins Drive, Iowa City, IA 52242, USA b Yale Center for Perinatal, Pediatric & Environmental Epidemiology One Church St, 6th Floor, New Haven, CT 06510, USA Accepted 15 November 2005

Abstract Background and Objective: To compare tracing and contact rates using alternative incentives in a computer-assisted telephone interview (CATI) survey among postpartum women. Methods: In a randomized trial of 1,061 postpartum women 18–49 years of age selected from four Iowa counties, we compared the effects of: (1) unconditional $5 telephone card incentive enclosed with the introductory letter followed by $25 incentive conditional upon successful telephone tracing, contact, and completion of CATI (Group 1, n 5 530) vs. (2) $30 incentive conditional upon subject completion of CATI (Group 2, n 5 531). Results: Overall telephone tracing and contact rates achieved were 67.8% and 66.6%, respectively. Tracing (70.2 vs. 65.4%, P 5 .09) and contact (68.5 vs. 64.8%, P 5 .26) rates were consistently higher among subjects assigned the combination of a conditional and an unconditional incentive. The combined incentive type had a greater impact on telephone tracing success rates for subjects on whom we could not initially locate an active telephone number (16.7 vs. 7.3%, P 5 .07) when compared to subjects for whom we found an active telephone number at the time of mailing the introductory letter (78.9 vs. 75.9%, P 5 .30). Conclusions: Combining conditional and unconditional recruitment incentives can facilitate telephone tracing efforts in surveys conducted among recently postpartum women. Ó 2006 Elsevier Inc. All rights reserved. Keywords: Computer-Assisted Telephone Interview; Contact rate; Incentive; Postpartum; Tracing rate; Survey

1. Introduction Recent studies have emphasized trends of decreasing tracing and contact rates in face-to-face interviews [1], telephone and computer-assisted telephone interviews (CATI) [2–11] as well as mail [10,12–24] and internet [25] surveys. Survey recruitment has become so problematic that a Summit, sponsored by the U.S. Census Bureau, was held in 2002. In their final report, the Summit Working Group recommended experimental assessment of unconditional incentives to motivate survey recruitment [26]. Most surveys to date have been conducted among subjects from the general population [1–4,7–10,18,20–22], while others have targeted low-income people [16], students [25], older women [24], and health professionals [6,12–15,17,19,23]. Few studies have addressed telephone tracing and contact rates among women of reproductive

* Corresponding author: Tel.: 319-384-5013; fax: 319-384-5004. E-mail address: [email protected] (A.F. Saftlas). 0895-4356/06/$ – see front matter Ó 2006 Elsevier Inc. All rights reserved. doi: 10.1016/j.jclinepi.2005.11.011

age or tested the effectiveness of using different incentives to promote recruitment in CATI surveys [27]. Survey topics targeted at women of reproductive age or postpartum women are often of sensitive nature. In addition, women who have recently delivered a new infant are less likely to commit time to a research study due to competing responsibilities. Thus, we expect that results from previous studies may not be applicable to this subgroup of women. Different methods have been used to enhance tracing and contact rates. Previous research has assessed monetary [1,6,10,12–18,20,21,23–25] and nonmonetary [13,16– 18,20,21,24,25] incentives or inducements. Some studies have compared a monetary incentive to no incentive [15,19]. Others have evaluated quantitatively different monetary incentives [6,12,16,20,23]. Still others have contrasted monetary and nonmonetary incentives [17,18,21,24,25], conditional and nonconditional incentives [14], as well as the impact of multiple incentives through a factorial design [18,25]. The aim of the present article is to compare tracing and contact rates achieved by two different incentive protocols

H. Beydoun et al. / Journal of Clinical Epidemiology 59 (2006) 732–738

among postpartum women. Specifically, the investigators designed this study to determine if tracing and contact rates could be improved by combining an unconditional incentive with a standard conditional incentive already in use, without altering the incentive-associated cost per completed interview. The purpose of an unconditional incentive is to increase the yield of women who would initiate contact with the project coordinating center, irrespective of whether or not they had a telephone. Study subjects were identified from birth certificate records belonging to residents of four Iowa counties, and invited to participate in an epidemiologic CATI survey. We hypothesized that women receiving an unconditional $5 telephone card enclosed with their introductory letter followed by $25 check conditional upon CATI completion (Group 1) would achieve improved rates over women randomized to receive $30 conditional upon CATI completion (Group 2).

2. Materials and methods 2.1. Study population A randomized trial of recruitment incentives was designed using a subsample of subjects from a large population-based case–control study. Data collection for the latter study is currently in the final stages. Women of reproductive age (18–49 years) who resided in four Iowa counties and delivered a singleton live birth between August and December 2002 were selected for recruitment in the current pilot study. Exclusion criteria included women less than 18 years at the time of delivery, multiple births, and those with diabetes mellitus, systemic lupus erythromatosus, or chronic renal disease. 2.2. Field operations Approval from the Institutional Review Board at the University of Iowa was received to conduct this methodologic randomized pilot study. A uniform protocol of field operations was implemented for all subjects selected for tracing. Initially, home addresses were obtained from birth certificate records and telephone numbers were traced using Web-based telephone directories, reverse directories, and directory assistance. Once a telephone number was ascertained, a ‘‘standard’’ introductory letter was sent. Otherwise, a ‘‘no-telephone’’ introductory letter was sent, which asked subjects to mail their telephone information to the research office or to contact the project’s toll-free number. If the introductory letter was returned by the post office with a forwarding address, the new address was used for retracing the telephone number and another letter was sent according to the protocol. Therefore, some of the subjects who initially received a standard introductory letter could end up with the disposition ‘‘untraceable’’ if an active telephone number is not found or provided by the subject. Other subjects who initially received a ‘‘no-telephone’’

733

letter may have an active telephone number identified by the end of the study period. It is worth noting that efforts were focused on identifying land line or regular telephone numbers. At the time the study was initiated, tracing could not be accomplished using cellular phone subscriber lists. In addition, potential participants were not asked whether they relied solely on a cellular telephone or if they also had a land line. Once an active telephone number was obtained, the subject was contacted by a project interviewer to: (1) obtain verbal consent for CATI participation and agreement to sign a medical record release form, (2) verify eligibility using a brief 3-min screening interview, and (3) set an appointment for the CATI. Telephone numbers were called at times that included mornings, afternoons, and evenings on weekdays. If the interviewer was not able to reach the subject within six call attempts, an ‘‘unable-to-contact’’ letter was sent and a maximum of four additional calls were attempted. Within the six call attempts, the interviewer was instructed to provide information to potential subjects about the study directly or through their answering machines. The length of the CATI interview averaged 50 min, with a range of 45 to 70 min. Monetary incentives of either $25 (Group 1) or $30 (Group 2) were mailed to all subjects who completed the CATI interview and agreed to provide researchers with their signed medical record release form. 2.3. Randomized trial of recruitment incentives A total of 1,061 women were randomly assigned to either one of two groups. Radomization was performed through the computerized database by assigning each record, at random, to two nearly equal sized groups. Group 1 (n 5 530) received an unconditional $5 telephone card included with the introductory letter, followed by a $25 check provided conditional upon completion of the CATI. Group 2 (n 5 531) was offered a $30 check conditional upon completion of the CATI. Prior to the trial, we had provided a $30 conditional incentive but were not meeting our projected tracing rate. Blinding of interviewers with respect to incentive assignment could not be achieved for logistical reasons. However, technicians who performed telephone tracing activities were blinded to the subjects’ incentive group assignment. Aside from the randomly assigned incentive type, other field aspects of the trial were equivalent for the two groups. Baseline data were obtained on all randomized subjects from the birth certificate files and linked to the detailed contact and call record database, which contains information on incentive assignment and field outcomes. Final tracing outcomes were: (1) active telephone traced and (2) no active telephone traced. Final contact outcomes were coded as one of the following: (1) agreed to be screened, (2) refused to be screened, and (3) no contact established because of inability to locate a telephone number or to reach selected subject.

734

H. Beydoun et al. / Journal of Clinical Epidemiology 59 (2006) 732–738

2.4. Sample size calculations The present analyses were conducted on a fixed sample size of 1,062 postpartum women selected from birth certificate files. Assuming equal allocation of subjects to treatment Groups 1 and 2, there would be 93% power to detect a 10% change (e.g., an increase in tracing or contact rate from 65 to 75%) at an alpha level of 0.05. 2.5. Statistical analysis All statistical analyses were conducted using the SAS system version 9.0. The Pearson chi-square test was used to compare Group 1 and Group 2 on baseline characteristics. Tracing rates were defined as the number of subjects on whom an active telephone number could be traced by the end of the study period divided by the total number of potentially enrolled subjects. Contact rates were calculated using formulas reported by Slattery et al. [28]. Independent samples tests comparing differences in tracing and contact rates among randomization groups were conducted using the normal approximation to the binomial distribution, assuming equal variances. Statistical significance and borderline statistical significance were determined at alpha levels of 0.05 and 0.1, respectively.

3. Results Table 1 presents baseline characteristics of subjects randomly assigned to either Group 1 or Group 2. As expected, no statistically significant differences (P ! .05) were observed between the two groups. The vast majority of subjects were between 20 and 34 years of age, with a mean age of 27 years at delivery. Nearly 33.3% had completed at least 4 years of college. Over 75% were non-Hispanic Whites and 66% were married. Almost 40% were nulliparous and 23% had a previous abortion. More than 99% received prenatal care and 17% had smoked during pregnancy, as recorded on the birth certificate. Based on data from the Centers for Disease Control and Prevention [29], our study sample is comparable to U.S. women (15–44 years of age) who gave birth in 2002 on various indicators including the percentage married or receiving prenatal care. On the other hand, 26% of all U.S. women had completed at least 4 years of college, 78% were non-Hispanic White and 11.4% smoked during pregnancy. Such disparities could be partially attributed to the exclusion of teenage pregnancies less 18 years of age from our study sample and more complete recording of smoking on Iowa birth certificates. In Figure 1, survey outcomes are described by incentive group and type of introductory letter received. Out of 1,061 potentially enrolled subjects, 900 (84.8%) (Group 1: 85.4% and Group 2: 84.5%) had a telephone number that could be initially traced and thereby received a ‘‘standard’’ introductory letter. On the other hand, 154 (15.2%) (Group 1: 13.6%

Table 1 Baseline characteristics of potentially enrolled subjects by incentive group Total Group 1a (n 5 1,061) (n 5 530) N (%) N (%) Age !20 81 20–24 289 25–29 313 30–34 245 351 133 Education Secondary or less 438 College (1–3 years) 269 College (41 years) 354 Race White, non-Hispanic 811 White, Hispanic 82 Black 96 Other 72 Marital status Married 699 Unmarried 364 Parity Nulliparous 425 Parous 635 Previous abortionsc No 800 Yes 251 Unknown 10 Number of prenatal care visits None 4 1–8 135 9–10 184 11–12 358 13–14 188 15–17 141 171 51 Tobacco during pregnancy Yes 174 No 886 Unknown 1 a b c

Group 2b (n 5 531) N (%)

P

(7.63) (27.24) (29.50) (23.09) (12.54)

40 135 169 117 69

(7.55) 41 (7.72) (25.47) 154 (29.00) (31.89) 144 (27.12) (22.08) 128 (24.11) (12.05) 64 (12.05) .41

(41.28) (25.35) (33.36)

215 (40.57) 223 (42.00) 130 (24.53) 139 (26.18) 185 (34.91) 169 (31.83) .56

(76.44) (7.73) (9.05) (6.79)

410 37 48 35

(65.15) (34.21)

350 (66.04) 349 (65.73) 180 (33.96) 182 (34.27) .91

(40.09) (59.91)

215 (40.57) 210 (39.62) 315 (59.43) 320 (60.38) .75

(75.40) (23.66) (0.94)

406 (76.60) 394 (74.20) 119 (22.45) 132 (24.86) 5 (0.94) 5 (0.94) .65

(0.38) (12.72) (17.34) (33.74) (17.72) (13.29) (4.81)

3 63 102 174 93 64 31

(16.40) (83.51) (0.09)

84 (15.85) 90 (16.95) 445 (83.96) 441 (83.05) 1 (0.19) 0 (0.00) .54

(77.36) 401 (75.52) (6.98) 45 (8.47) (9.06) 48 (9.04) (6.60) 37 (6.97) .82

(0.57) 1 (0.19) (11.89) 72 (13.56) (19.25) 82 (15.44) (32.83) 184 (34.65) (17.55) 95 (17.89) (12.08) 77 (14.50) (5.85) 20 (3.77) .26

Group 1 received unconditional $5 telephone card and conditional $25. Group 2 received conditional $30. Previous abortions, before 20 weeks of gestation.

and Group 2: 15.5%) of the subjects initially did not have a traceable telephone number and were therefore sent a ‘‘no-telephone’’introductory letter. Telephone contact was established with 618 (58.2%) postpartum women (Group 1: 60.5% and Group 2: 55.9%). Once contact with a potential subject was established, more than 80% of potentially eligible women agreed to be screened according to prespecified eligibility criteria. It is worth noting that three subjects (two in Group 1 and one in Group 2) were successfully contacted even though their telephone number could not be traced. These subjects had taken the initiative to contact the coordinating center using the 1-800- number provided in the body of the introductory letter.

H. Beydoun et al. / Journal of Clinical Epidemiology 59 (2006) 732–738

735

Incentive study sample (N=1,061) Random assignment

Group 2b N=531

Group 1a N=530 Standard intro letterc

No phone intro letter

Standard intro letter

No phone intro letter

N=452 Traced: 357; Not traced: 95

N=72 Traced: 12; Not traced: 60

c

N=448 Traced: 340; Not Traced: 108

c

N=82 Traced: 6; Not traced: 76

c

Successful contact

Successful contact

Successful contact

Successful contact

N=307

N=14

N=290

N=7

Agree to be screened

Agree to be screened

Agree to be screened

Agree to be screened

N=277

N=11

N=240

N=7

aGroup 1 received unconditional $5 telephone card and conditional $25; bGroup 2 received conditional $30; cIn seven subjects, no data were available on initial type of introductory letter.

Fig. 1. Pilot study summary of field operation outcomes.

A stratified analysis of tracing and contact rates among the randomization groups by introductory letter type is presented in Table 2. The overall tracing and contact rates achieved were 67.8 and 66.6%, respectively. Tracing (70.2 vs. 65.4%, P 5 .09) and contact (68.5 vs. 64.8%, P 5 .26) rates were consistently higher among subjects randomized to receive a combination of conditional and unconditional incentives. The differences in tracing rates were of borderline significance (P ! .1). Furthermore, the combined incentive had a substantially improved effect

on tracing rates for subjects who initially did not have an active telephone number (16.7 vs. 7.3%, P 5 .07), but had little effect on tracing rates among those who initially had an active telephone number (78.9% vs. 75.9%, P 5 .3). Despite modest achievements in improving tracing and contact rates, addition of a telephone card as an unconditional incentive was instrumental in reaching a subgroup of socioeconomically disadvantaged women for whom a telephone number could not be traced initially. Further analyses confirmed that this subgroup had a greater

Table 2 Tracing and contact rates by incentive group and type of introductory letter Tracing ratea c

Overall Group 1d Group 2e Standard introductory letter Group 1 Group 2 No telephone introductory letter Group 1 Group 2

n

%

1061 530 531 900 452 448 154 72 82

67.8 70.2 65.4 77.4 78.9 75.9 11.7 16.7 7.3

Contact rateb P

.092

.267

.072

n

%

1061 530 531 900 452 448 154 72 82

66.6 68.5 64.8 73.1 74.3 71.8 31.2 36.1 26.8

P

.255

.197

.107

Abbreviations: T, active telephone traced; NT, no active telephone traced; I, interviewed individuals; P, partially interviewed individuals; R, refusals; NE, not eligible for study criteria; NCP, no contact because of inability to locate a number or unable to reach selected subject. a Tracing rate 5 T/(T1NT). b Contact rate 5 (I1P1R1NE)/(I1P1R1NCP). c In seven subjects, no data was available on initial type of introductory letter. d Group 1 received unconditional $5 telephone card and conditional $25. e Group 2 received conditional $30.

736

H. Beydoun et al. / Journal of Clinical Epidemiology 59 (2006) 732–738

percentage of young, unmarried, less educated, non-White women who reported a higher smoking rate during pregnancy, and had fewer prenatal care visits (Table 3). 4. Discussion Few studies to date have assessed the effectiveness of different incentives in improving recruitment to a CATI survey [6,30] among postpartum women, who constitute a distinct population from health professionals and other special populations, and might respond differently to similar recruitment strategies [24]. In the current randomized controlled trial, the group of women randomly assigned to receive an unconditional $5 telephone card in addition to a $25 conditional incentive were traced and contacted at higher rates than women who were randomized to $30 conditional incentive. Among women who initially received

a ‘‘no-telephone’’ introductory letter, the $5 telephone card was particularly instrumental in achieving higher tracing and contact rates. Several factors have played a role in determining the tracing and contact rates in our study. These include characteristics of subjects (postpartum women), survey method used (CATI), efforts spent on tracing and recruitment of subjects (field operations), in addition to qualitative and quantitative characteristics of the incentives provided (check and telephone card). Often, the survey method is predetermined by the target population of interest, especially when subjects are health professionals [6,12– 15,17,19,23]. In other situations, a choice of different survey methods, including telephone [2–4,8], mailed [16,20], or direct interview [1] surveys or a combination of these [5,10] could be applied. Telephone and CATI surveys have many advantages over other methods of survey

Table 3 Baseline characteristics of potentially enrolled subjects by type of introductory letter received initially

Age !20 20–24 25–29 30–34 351 Education Secondary or less College (1–3 years) College (41 years) Race White, non-Hispanic White, Hispanic Black Other Marital status Married Unmarried Parity Nulliparous Parous Previous abortionsb No Yes Unknown Number of prenatal care visits None 1–8 9–10 11–12 13–14 15–17 171 Tobacco during pregnancy Yes No a b

Totala (n 5 1,054) N (%)

Standard introductory letter (n 5 900) N (%)

81 285 312 244 132

52 233 268 228 119

(7.69) (27.04) (29.60) (23.15) (12.52)

(5.78) (25.89) (29.78) (25.33) (13.22)

433 (41.08) 269 (25.52) 352 (33.40)

323 (35.89) 242 (26.89) 335 (37.22)

804 82 96 72

722 59 60 59

(76.28) (7.78) (9.11) (6.83)

(80.22) (6.56) (6.67) (6.56)

No-telephone introductory letter (n 5 154) N (%) 29 52 44 16 13

P

(18.83) (33.77) (28.57) (10.39) (8.44)

! .0001

110 (71.43) 27 (17.53) 17 (11.04)

! .0001

82 23 36 13

(53.25) (14.94) (23.38) (8.44)

! .0001

695 (65.94) 359 (34.06)

642 (71.33) 258 (28.67)

53 (34.42) 101 (65.58)

! .0001

423 (40.17) 630 (59.83)

361 (40.16) 538 (59.84)

62 (40.26) 92 (59.74)

.98

794 (75.33) 250 (23.72) 10 (0.95)

686 (76.22) 205 (22.78) 9 (1.00)

108 (70.13) 45 (29.22) 1 (0.65)

.21

4 133 183 355 188 141 50

3 100 157 304 162 129 45

(0.38) (12.62) (17.36) (33.68) (17.84) (13.38) (4.74)

171 (16.22) 883 (83.78)

(0.33) (11.11) (17.44) (33.78) (18.00) (14.33) (5.00)

129 (14.33) 771 (85.67)

In seven subjects, no data were available on initial type of introductory letter. Previous abortions, before 20 weeks of gestation.

1 33 26 51 26 12 5

(0.65) (21.43) (16.88) (33.12) (16.88) (7.79) (3.25)

.0107

42 (27.27) 112 (72.73)

! .0001

H. Beydoun et al. / Journal of Clinical Epidemiology 59 (2006) 732–738

administration. However, some major obstacles to achieving adequate recruitment have been encountered recently in telephone and CATI surveys. Some issues in telephone tracing and contact that have become particularly problematic in recent years include increased usage of unlisted cellular telephone numbers, less tolerance for interruptions of private life by telemarketers, and greater usage of technology that screens incoming telephone calls [31,32]. To date, most studies that have assessed the impact of monetary or nonmonetary incentives on recruitment success were limited to mailed surveys of health professionals [12–15,17,19,23,30]. These studies suggested varying improvements in subject recruitment by type of incentive, with monetary incentives being more successful than other incentive types. In the current pilot study of recruitment incentives, moderate overall tracing (67.8%) and contact (66.6%) rates were achieved, irrespective of inducement strategy. The telephone card incentive played a modest role in improving these rates, but was highly instrumental in motivating subjects whose telephone number could not be initially traced. Findings from the pilot study have been translated into further changes in the survey protocol, among which was the administration of the more effective incentive type (unconditional $5 telephone card enclosed with their introductory letter followed by $25 check conditional upon CATI completion) to all subsequently identified postpartum women and implementation of Saturday recruiting and interviewing. In conclusion, combining conditional and unconditional recruitment incentives can facilitate telephone tracing efforts in surveys conducted among postpartum women. Acknowledgments This study was supported by grant RO1-HD39753-01 from the National Institute of Child Health and Human Development. We would like to thank all tracing and interviewing staff who worked on this pilot project. References [1] Pavlik VN, Hyman DJ, Vallbona C, Dunn JK, Louis K, Dewey CM, Wieck L, Toronjo C. Response rates to random digit dialing for recruiting participants to an onsite health study. Public Health Rep 1996;111:444–50. [2] Anie KA, Jones PW, Hilton SR, Anderson HR. A computer-assisted telephone interview technique for assessment of asthma morbidity and drug use in adult asthma. J Clin Epidemiol 1996;49:653–6. [3] Blyth FM, March LM, Shellard D, Cousins MJ. The experience of using random digit dialing methods in a population-based chronic pain study. Aust N Z J Public Health 2002;26:511–4. [4] Corkrey R, Parkinson L. A comparison of four computer-based telephone interviewing methods: getting answers to sensitive questions. Behav Res Methods Instrum Comput 2002;34:354–63. [5] Groves RM, Mathiowetz NA. Computer assisted telephone interviewing: effects on interviewers and respondents. Public Opin Q 1984;48: 356–69.

737

[6] Gunn WJ, Rhodes IN. Physician response rates to a telephone survey: effects of monetary incentive level. Public Opin Q 1981;45:109–15. [7] Hornik J, Zaig T, Shadmon D, Barbash GI. Comparison of three inducement techniques to improve compliance in a health survey conducted by telephone. Public Health Rep 1990;105:524–9. [8] Ketola E, Klockars M. Computer-assisted telephone interview (CATI) in primary care. Fam Pract 1999;16:179–83. [9] Slade GD, Brennan D, Spencer AJ. Methodological aspects of a computer-assisted telephone interview survey of oral health. Aust Dent J 1995;40:306–10. [10] Wang PS, Beck AL, McKenas DK, Meneades LM, Pronk NP, Saylor JS, Simon GE, Walters EE, Kessler RC. Effects of efforts to increase response rates on a workplace chronic condition screening survey. Med Care 2002;40:752–60. [11] Wilkins D, Casswell S, Barnes HM, Pledger M. A pilot study of a computer-assisted cell-phone interview (CACI) methodology to survey respondents in households without telephones about alcohol use. Drug Alcohol Rev 2003;22:221–5. [12] Asch DA, Christakis NA, Ubel PA. Conducting physician mail surveys on a limited budget. A randomized trial comparing $2 bill versus $5 bill incentives. Med Care 1998;36:95–9. [13] Deehan A, Templeton L, Taylor C, Drummond C, Strang J. The effect of cash and other financial inducements on the responser of general practitioners in a national postal study. Br J Gen Pract 1997;47: 87–90. [14] Delnevo CD, Abatemarco DJ, Steinberg MB. Physician response rates to a mail survey by specialty and timing of incentive. Am J Prev Med 2004;26:234–6. [15] Everett SA, Price JH, Bedell AW, Telljohann SK. The effect of a monetary incentive in increasing the return rate of a survey to family physicians. Eval Health Prof 1997;20:207–14. [16] Gibson PJ, Keopsell TD, Diehr P, Hale C. Increasing response rates for mailed surveys of Medicaid clients and other low-income populations. Am J Epidemiol 1999;149:1057–62. [17] Oden L, Price JH. Effects of a small monetary incentive and followup mailings on rates of a survey to nurse practitioners. Psychol Rep 1999;85:1154–6. [18] Perneger TV, Etter JF, Rougemont A. Randomized trial of use of a monetary incentive and a reminder to increase the response rate to a mailed health survey. Am J Epidemiol 1993;138:714–22. [19] Russell ML, Mustasingwa DR, Verhoef MJ, Injeyan HS. Effect of monetary incentive on chiropractors’ response rate and time to respond to a mail survey. J Clin Epidemiol 2003;56:1027–8. [20] Shaw MJ, Beebe TJ, Jensen HL, Adlis SA. The use of monetary incentives in a community survey: impact on response rates, data quality, and cost. Health Serv Res 2001;35:1339–46. [21] Spry VM, Hovell MF, Sallis JG, Hofsteter CR, Elder JP, Molgaard CA. Recruiting survey respondents to mailed surveys: controlled trials of incentives and prompts. Am J Epidemiol 1989; 130:166–72. [22] Stockbridge H, Hardy RI, Glueck CJ. Public cholesterol screening: motivation for participation, follow-up outcome, self-knowledge, and coronary heart disease risk factor intervention. J Lab Clin Med 1989;114:142–51. [23] VanGeest JB, Wynia MK, Cummins DS, Wilson IB. Effects of different monetary incentives on the return rate of a national mail survey of physicians. Med Care 2001;39:197–201. [24] Whiteman MK, Langenberg P, Kjerulff K, McCarter R, Flaws JA. A randomized trial of incentives to improve response rates to a mailed women’s health questionnaire. J Womens Health 2003;12:821–8. [25] Kypri K, Gallaguer SJ. Incentives to increase participation in an internet survey of alcohol use: a controlled experiment. Alcohol Alcohol 2003;38:437–41. [26] Salvucci S, Wenck S, Hamsher S, Bates N. Response rate summitdNational Health Interview & Consumer Expenditure Quarterly Surveys. U.S. Census Bureau and the Interagency Household Survey Nonresponse Group; 2002.

738

H. Beydoun et al. / Journal of Clinical Epidemiology 59 (2006) 732–738

[27] Brown JE, Jacobs DR Jr, Barosso GM, Potter JD, Hannan PJ, Kopher RA, Rourke MJ, Hartman TJ, Hase K. Recruitment, retention and characteristics of women in a prospective study of preconceptional risks of reproductive outcomes: experience of the Diana Project. Paediatr Perinat Epidemiol 1997;11:345–58. [28] Slattery ML, Edwards SL, Caan BJ, Kerber RA, Potter JD. Response rates among control subjects in case–control studies. Ann Epidemiol 1995;5:245–9. [29] Centers for Disease Control and Prevention. National vital statistics reports. Births: final data for 2002. December 17, 2003.

[30] Edwards P, Cooper C, Roberts R, Frost C. Meta-analysis of randomized trials of monetary incentives and response to mailed questionnaires. J Epidemiol Community Health 2005;59:987–99. [31] Nayak MB, Kaskutas LA. Risky drinking and alcohol use patterns in a national sample of women of childbearing age. Addiction 2004;99: 1393–402. [32] Midanik LT, Greenfield TK. Defining ‘‘current drinkers’’ in national surveys: results of the 2000 National Alcohol Survey. Addition 2003;98:517–22.