A comparison of costs and data quality of three health survey methods: mail, telephone and personal home interview

Is Telephone Screening Feasible? Accuracy and Cost-Effectiveness of Identifying People Medically Eligible for Home- and Community-Based Services

PURPOSE

To determine the accuracy of a telephone-screening system to identify persons eligible for home- and community-based long-term care.

DESIGN AND METHODS

Data from Michigan telephone screens were compared to data from in-person assessments using the Minimum Data Set for Home Care (MDS-HC). Weighted kappa statistics measured the level of agreement between the two assessments.

RESULTS

Overall, recommendations based on the telephone screen produced a marginal match compared to recommendations based on in-person assessment. "False positives" (individuals scoring as more impaired on the telephone screen than in person) occurred in 27% of all cases, while "false negatives" (individuals scoring as less impaired on the telephone screen) only occurred among 6% of the callers. Neither individual screen questions, source of information, location of the individual, timing between screen and assessment, nor temporal changes accounted for mismatches. Telephone screens resulted in an 11% savings over the cost of providing in-person assessments to all program seekers.

IMPLICATIONS

The telephone screen has utility as a broad targeting mechanism that allows agencies to avoid costly in-person assessments for all program seekers. Evidence does not support use of the telephone screen alone to determine either medical eligibility or a specific level of care.

Surveying Minorities with Limited-English Proficiency

BACKGROUND

Little is known about how modes of survey administration affect response rates and data quality among populations with limited-English proficiency (LEP). Asian Americans are a rapidly growing minority group with large numbers of LEP immigrants.

OBJECTIVE

We sought to compare the response rates and data quality of interviewer-administered telephone and self-administered mail surveys among LEP Asian Americans.

DESIGN

This was a randomized, cross-sectional study using a 78-item survey about quality of medical care that was given to Vietnamese, Mandarin, or Cantonese Chinese patients in their native language.

MEASURES

We examined response rates and missing data by mode of survey and language groups. To examine nonresponse bias, we compared the sociodemographic characteristics of respondents and nonrespondents. To assess response patterns, we compared the internal-consistency reliability coefficients across modes and language groups.

RESULTS

We achieved an overall response rate of 67% (322 responses of 479 patients surveyed). A higher response rate was achieved by phone interviews (75%) as compared with mail surveys with telephone reminder calls (59%). There were no significant differences in response rates by language group. The mean number of missing item for the mail mode was 4.14 versus 1.67 for the phone mode (P< or =0.000). There were no significant differences in missing data among the language groups and no significant differences in scale reliability coefficients by modes or language groups.

CONCLUSIONS

Telephone interviews and mail surveys with phone reminder calls are feasible options to survey LEP Chinese and Vietnamese Americans. These methods may be less costly and labor-intensive ways to include LEP minorities in research.

Quality of life after lung cancer resection

Lung cancer continues to be the most common cancer in the world, with the highest cancer mortality rate by far. Although resection remains the treatment of choice in early-stage NSCLC, the prognosis remains grim even after surgical treatment. In a patient population with such a high mortality rate, evaluation and preservation of QOL after treatment is imperative. Early-stage lung cancer patients already have significantly lower QOL when compared with the normal population before surgical treatment, with significant impairment in physical and emotional functioning. Lung cancer resection causes further deterioration of QOL, especially in the first 3 to 6 months after surgery. While some studies suggest that QOL returns to baseline levels at 6 to 9 months postoperatively, others report that QOL is still significantly impaired at 6 months and 1 year after surgery. Although prospective studies analyzing long-term postoperative QOL are not available, retrospective data suggest that long-term survivors after lung cancer surgery enjoy good QOL despite impaired physical functioning. QOL studies on VATS lung cancer resection are extremely limited. More prospective, longitudinal studies with larger study populations and longer follow-up periods are needed to portray the course of QOL in lung cancer patients more accurately and to improve postoperative care. Furthermore, comparative studies between VATS and the standard thoracic incisions (including QOL assessments) must be performed to guide clinical decision making regarding selection of optimal access modality for performing lung cancer resection.

Measurement strategies designed and tested in the Veterans Health Study

The goal of collecting a large number of health status variables from a population of elderly respondents with high comorbidity offers several challenges and opportunities that are described in this article. The data collection strategies used in the Veterans Health Study (VHS) are discussed, and we detail the development and item content of the questionnaires used. The VHS included various modes of administration, in-person interviews, interviewer and self-administered questionnaires, proxy interviews, and administrative databases to ensure a more complete assessment of health status. Included is a discussion of staffing and the training methods developed. The article concludes with lessons learned that may be of interest to other researchers conducting similar health outcomes studies.

Comparison of the Roland-Morris Disability Questionnaire and generic health status measures: a population-based study of workers' compensation back injury claimants

STUDY DESIGN

Workers' compensation back injury claimants completed baseline and follow-up telephone interviews in a prospective population-based cohort study.

OBJECTIVE

To compare the Roland-Morris Disability Questionnaire (RDQ) to widely used generic health status measures in a sample of workers with recent work-related back injuries in terms of validity, reliability, responsiveness to change, and floor and ceiling effects.

SUMMARY OF BACKGROUND DATA

Little research has directly compared the validity and responsiveness of the RDQ to that of the Short-Form 12 or Short-Form 36 health status measures among individuals with back pain. Furthermore, there is little information concerning the validity, reliability, and responsiveness of the RDQ as a measure of functional outcomes for workers with back injuries.

METHODS

Approximately 8 weeks (median) after filing low back injury claims, 309 workers completed the RDQ, Short-Form 12, and Short-Form 36 scales and gave information about their work status in computer-assisted telephone interviews. An average of 5 months later, 284 workers (91.9%) completed the measures again.

RESULTS

The RDQ demonstrated excellent internal consistency and validity through correlations with other measures of physical functioning, ability to discriminate between those working and those not working, and much more responsiveness to change than the Short-Form 12 and Short-Form 36 scales. However, 15% of the sample did not answer one or more RDQ items.

CONCLUSIONS

The RDQ is a valid measure of physical disability among workers with back injuries. Its greater responsiveness to change suggests its superiority to the Short-Form 12 and Short-Form 36 as an outcome measure in this population.

Late effects of childhood cancer therapy

This study reviews the common long-term sequalae of childhood cancer and its therapy. It discusses the clinical and research challenges posed by such late effects. The authors address related topics of late effects research and clinical care, methodological issues, barriers and directions for the future.

The effect of data collection mode on smoking attitudes and behavior in young African American and Latina women. Face-to-face interview versus self-administered questionnaires

Evaluating smoking prevention and cessation programs requires valid data collection. This study examined two survey modes--face-to-face (FTF) interview and self-administered questionnaire (SAQ)--comparing response rates, sample characteristics, data quality, and response effects. From two family planning clinics, 601 female Latina and African American clients ages 12 to 21 were recruited and randomized to either group. Results reveal that neither mode is superior to the other. The SAQ may therefore be preferable for this population, despite its higher rate of incompletes, because it yields results similar to the FTF yet is more cost effective and less disruptive to clinic routines.

National health surveys by mail or home interview: effects on response

STUDY OBJECTIVE

To study the effect of using a mail questionnaire or home interviews on the size and the selectivity of response to national health surveys.

DESIGN

The interview survey and the mail survey were both carried out in the same country (the Netherlands) using the same sample frame, the same study period (1998) and collected partly the same data on demographic, socioeconomic and health characteristics.

SETTING

The Netherlands.

PARTICIPANTS

Dutch non-institutionalised inhabitants aged 25 years and over.

MAIN RESULTS

Response to the mail survey was lower (46.9%, n=3664) than to the interview survey (58.5%, n=6061). The mail survey gave higher response rates for women and lower response rates for persons with lower levels of education. Respondents to the mail survey reported lower rates of smoking but a slightly worse health status and higher figures on the use of health care services. No differences by method of data collection were found for age, marital status, region, household composition, work status and categories of body mass index.

CONCLUSION

Although the response of the mail survey was lower than the home interview survey, respondents showed generally small differences, with exception of level of education.

Validity of a telephone-administered 24-hour dietary recall in telephone and non-telephone households in the rural Lower Mississippi Delta region

OBJECTIVE

To determine if 24-hour dietary recall data are influenced by whether data are collected by telephone or face-to-face interviews in telephone and non-telephone households.

DESIGN

Dual sampling frame of telephone and non-telephone households. In telephone households, participants completed a 24-hour dietary recall either by face-to-face interview or telephone interview. In non-telephone households, participants completed a 24-hour dietary recall either by face-to-face interview or by using a cellular telephone provided by a field interviewer.

SUBJECTS/SETTING

Four hundred nine participants from the rural Delta region of Arkansas, Louisiana, and Mississippi.

MAIN OUTCOME MEASURES

Mean energy and protein intakes.

STATISTICAL ANALYSES PERFORMED

Comparison of telephone and non-telephone households, controlling for type of interview, and comparison of telephone and face-to-face interviews in each household type using unpaired t tests and linear regression, adjusting for gender, age, and body mass index.

RESULTS

Mean differences between telephone and face-to-face interviews for telephone households were -171 kcal (P = 0.1) and -6.9 g protein (P = 0.2), and for non-telephone households -143 kcal (P = 0.6) and 0.4 g protein (P = 1.0). Mean differences between telephone and non-telephone households for telephone interviews were 0 kcal (P = 1.0) and -0.9 g protein (P = 0.9), and for face-to-face interviews 28 kcal (P = 0.9) and 6.4 g protein (P = 0.5). Findings persisted when adjusted for gender, age, and body mass index. No statistically significant differences were detected for mean energy or protein intake between telephone and face-to-face interviews or between telephone and non-telephone households.

APPLICATIONS/CONCLUSIONS

These data provide support that telephone surveys adequately describe energy and protein intakes for a rural, low-income population.

Evaluating the administration of an office-based health survey in a primary care practice

Self-administered waiting room questionnaires are popular tools for gathering health information from patients, but these data cannot be used for research purposes without confirming adequate sampling of the practice population and assessing the completeness and accuracy of patients' responses. Long-term data collection also requires avoiding an imposition on clinic operations. We developed a protocol to test these questions in a 9-week pilot study of 884 survey-eligible patients visiting a family practice clinic. We found an adequate proportion of eligible patients were approached (74%) and participated (89%), they provided relatively complete (82-98%) and accurate responses, and the impact on office operations was minimal (<2 min of staff time per participant). Some demographic differences in participation and survey item completion were identified. A systematic process for testing survey performance allowed us to not only document these findings, but also to rapidly identify problems and introduce solutions while the survey was in progress.

Issues associated with the application to veterinarians of a mailed questionnaire regarding lower respiratory-tract disease in racehorses

This study evaluated a questionnaire investigating the attitudes and behaviours of veterinarians regarding the cause, diagnosis and treatment of lower respiratory-tract disease in racehorses. The questionnaire was sent to all non-student members (648) of the Australian Equine Veterinary Association: two mailings and a single telephone contact (each separated by four weeks). Subsequent phases were only administered to those who had not responded to earlier phases. In total, 467 (72.1%) of the 648 mailed questionnaires were returned. Of these, 354 were usable. The remaining 113 respondents gave various reasons for not completing the questionnaire; the most common (68%) was that horses were not a component of their practice. Those respondents working primarily with horses required fewer phases to return the questionnaire. Although deviating from previously described questionnaire designs, the described protocol provided a reasonable response rate.

Accuracy of proband reported family history: the NHLBI Family Heart Study (FHS)

Proband-reported family histories are widely used in research and counseling, yet little is known about the validity of family history reporting. The Family Heart Study (FHS), a population-based study of familial cardiovascular disease, gathered family history information from 3,020 middle-aged probands in four U.S. communities. Probands reported on the history of coronary heart disease (CHD), diabetes, hypertension, and asthma among a total of 10,316 living relatives (9,186 siblings, 1,130 parents) and 2,685 spouses. Questionnaires were returned by 6,672 siblings, 901 parents, and 2,347 spouses, yielding response rates of 73, 79, and 87%, respectively. Utilizing the relatives' self-report as the standard, sensitivity of the proband report on their spouse, parent, and sibling was 87, 85, and 81% for CHD, 83, 87, and 72% for diabetes, 77, 76, and 56% for hypertension, and 66, 53, and 39% for asthma, respectively. Most specificity values were above 90%. Analyses using generalized estimating equations (GEE) were performed to evaluate differences in proband accuracy based on the proband's age, gender, disease state, center, and ethnicity. In multivariate models, age, gender, and disease status were significantly associated with the accuracy of proband's report of sibling disease history, but had little effect on the accuracy of their report on spouses or parents. In general, older probands were significantly less accurate reporters of disease than younger probands. These results demonstrate that CHD family history can be captured effectively based on proband reports, but suggest that additional family contacts may be helpful when working with older probands or with chronic diseases that have few recognized medical events or procedures.

Evaluation of the short-form 36-item questionnaire to measure health-related quality of life in patients with idiopathic pulmonary fibrosis

OBJECTIVE

To validate the use of the 36-item short-form questionnaire (SF-36) for measuring health-related quality of life (HRQL) in patients with idiopathic pulmonary fibrosis (IPF).

DESIGN

: Observational data at a single point in time.

SETTING

: A specialized outpatient respiratory clinic.

PARTICIPANTS

Thirty-four patients (mean +/- SE age, 58.29 +/- 1.87 years) with IPF and no significant comorbidity. A matched control group for HRQL measurements was composed of 34 normal subjects (mean age, 58.00 +/- 1.89 years).

MEASUREMENTS AND RESULTS

Dyspnea was measured by the baseline dyspnea index (BDI). Respiratory function evaluation included FVC, FEV(1), and resting arterial blood gases. IPF patients showed a mean BDI score of 5.21 +/- 0.46. The mean FVC and FEV(1) values were 62.41 +/- 2.96% and 66.41 +/- 3.33%, respectively. The mean PaO(2) was 67 +/- 2.51 mm Hg, and the mean PaCO(2) was 37 +/- 1. 05 mm Hg. Patients scored significantly worse than control subjects with respect to the SF-36 domains of physical functioning, physical role, general health perceptions, vitality, social functioning, emotional role, and mental health index. BDI scores were significantly correlated with five SF-36 components, and FVC and FEV(1) were significantly correlated with two SF-36 components. Significant negative correlations were found between arterial pH and four SF-36 domains.

CONCLUSIONS

Patients with IPF have a significant impairment of HRQL in both physical and psychological functioning. Dyspnea is the most important factor influencing the quality of life in these subjects. The SF-36 questionnaire is a valid instrument to evaluate HRQL in IPF patients.

Telephone follow-up in a randomized controlled trial in a less developed country: feasibility, validity and representativeness

Telephone follow-up would offer considerable advantages over other modes of follow-up in less developed countries, if it were feasible and the findings were valid and generalizable. Telephone follow-up was assessed in the context of a randomized controlled trial of chest radiography in South African children. Hospital-based clinical outcomes were measured from hospital records, and also by telephone, in a subset of the same patients who offered a contact telephone number. Of 398 subjects offering a telephone number 308 (77.4%) were followed to recovery or for 28 days. Kappa statistics for a subsequent hospital visit, hospital admission, and chest radiograph were 0.88, 0.83, and 0.56, respectively. The effect of chest radiography did not differ significantly in participants accessible and not accessible by telephone. Telephone follow-up was feasible and produced valid and generalizable results at low cost.

Snowball sampling by mail: application to a survey of smokers in the general population

BACKGROUND

In a series of surveys intended for current and former smokers but sent to a random sample of the general population, we asked never smokers and smokers who did not wish to participate to transmit the questionnaire to any ever smoker they knew. We compared participants who received the questionnaire directly from us (original participants) to participants who received it from an addressee (secondary participants).

METHODS

Questionnaires on smoking were mailed to 3300 residents of Geneva (Switzerland) in 1997, and returned by 1167 people (35%).

RESULTS

The final sample consisted of similar numbers of original participants (n = 578, primary response rate = 18% of total sample, or about 46% of ever smokers) and secondary participants (n = 566). Original participants were 1.7 years older than secondary participants (P = 0.03) and were more likely to be men (50% versus 43%, P = 0.009). Proportions of current smokers, stages of change, confidence in ability to quit smoking, cigarettes per day and attempts to quit smoking were similar in the two groups. Secondary participants had lower self-efficacy scores (-0.30 standard deviation (SD) units, P < 0.03), and they derived more pleasure from smoking (+0.25 SD units, P = 0.04). Among ex-smokers, direct participants were less active than secondary participants in coping with the temptation to smoke (-0.58 SD units, P = 0.002). Associations between smoking-related variables were similar in original and secondary participants.

CONCLUSION

Allowing non-eligible addressees to transmit the questionnaire to someone else doubled the response rate, produced moderate bias on some variables only and had no detectable impact on associations between smoking-related variables.

Routine follow-up office visits after total joint replacement: do asymptomatic patients wish to comply?

A total of 100 patients presenting for routine office follow-up after total hip or knee arthroplasty completed questionnaires evaluating whether they preferred to come to the office for routine follow-up evaluation or whether they would have preferred an evaluation without an office visit. Of 100 patients, 45 would have preferred not to come into the office for a routine evaluation. They were content to mail completed questionnaires and radiographs to their physicians. The other 55 patients preferred office visits. These 2 groups were comparable for age, sex, height, weight, and number of surgeries (P > .11) Preoperative and postoperative scores were similar between the 2 groups (P > .39). None of the patients that would have preferred not to come in to the office believed that quality of care would be compromised. A significant number (45%) of patients would prefer not to come to the office because of the wages saved and time spared. Routine office visits may be eliminated for these patients through the use of health outcome devices, such as the SF-36, along with routine radiographs. The potential to decrease healthcare costs and increase patient satisfaction warrants the identification of these patients. Assessment of the effect on quality of care with elimination of routine follow-up visits requires further study.

The use of telephone interview methodology to obtain 24-hour dietary recalls

OBJECTIVE

To compare 24-hour dietary recalls collected over the telephone to in-person recalls collected in the 1994-1996 Continuing Survey of Food Intakes by Individuals (CSFII).

DESIGN

Trained interviewers collected 24-hour dietary recalls over the telephone using the multiple-pass approach. These results were compared to in-person interviews from a pooled subsample of CSFII respondents.

SUBJECTS/SETTING

List-assisted random-digit dialing was used to identify 700 women between the ages of 20 and 49 years. One eligible woman per household was selected to participate.

STATISTICAL ANALYSES

Approximate t tests to examine differences in average nutrient and energy intakes were conducted on weighted data.

RESULTS

The reported intakes of most nutrients in the current 24-hour dietary recalls collected over the telephone were significantly higher than those reported in the 1994 and 1995 CSFII, but there were no significant differences between the telephone survey and 1996 CSFII results. The 24-hour dietary recalls collected over the telephone yielded consistently greater mean nutrient intake per respondent compared with a comparable pooled subsample from the 1994, 1995, and 1996 CSFII. Generally, no significant differences were found in the food group data between the telephone survey and the CSFII survey. Mean dietary intakes reported by the comparable CSFII subsample increased from 1994 to 1996.

APPLICATIONS

Collecting 24-hour dietary recalls over the telephone is a practical and valid data collection tool for use in national food consumption surveys.

A population comparison of participants and nonparticipants in a health survey

OBJECTIVES

This study examined the characteristics of Air Force recruits willing to take part in a health survey vs those unwilling to participate.

METHODS

US Air Force recruits undergoing basic military training (n = 32,144) were surveyed regarding demographic and health variables.

RESULTS

Respondents indicating an unwillingness to participate in a health survey reported less healthy lifestyles than those willing to participate. Prediction equations modeling the characteristics of those engaging in 4 risky behaviors were nearly identical regardless of whether those refusing to participate were included.

CONCLUSIONS

Results suggest that, despite some low estimates of health behaviors due to response bias, relationships between most risk factors are generally unaffected by those not responding to health surveys.

Screening seniors for risk of functional decline: results of a survey in family practice

OBJECTIVE

To measure functional status, determine risk of functional decline and assess consistency between responses and standardized instruments.

DESIGN

A mailed survey which measured functional impairment, recent hospitalization and bereavement. A positive response on at least one of these factors indicated that the individual was "at risk" for functional decline. A random sample (n = 73) of "at risk" subjects (specifically, family practice patients aged 70 and older) were assessed by a nurse.

RESULTS

The response rate was 89% (369/415), 59% of seniors were female and the mean age was 77.1 (SD = 5.5) years. Self-reported risk, based on activities of daily living (ADLs), was associated with impairment in at least one basic ADL (p < 0.0005) using a standardized instrument. The positive predictive value of the survey for ADL impairment was 65%.

CONCLUSION

Response to a mailed survey was high and self-reported ADL risks were consistent with findings from standardized assessment tools.

An examination of self- and telephone-administered modes of administration for the Australian SF-36

The impact of administration mode on health-related quality of life measurement instruments has received little attention, especially for the SF-36. General community members were randomly selected to complete the Australian version of the SF-36 Health Survey using either telephone or mail modes. Modes were compared across a number of indices: data collection costs; consent rates and non-consent bias; data quality (completeness of data and internal consistency reliability); and response effects. Data collection costs were lower for the telephone mode. A significantly higher consent rate was achieved with the telephone mode. Those who were younger were more likely to refuse to participate when the mail mode was adopted, while older people were more likely not to consent to the telephone mode. The rate of missing responses was higher for the mail mode, while significant differences were found between modes in internal consistency reliability estimates. Health ratings were more favorable for the telephone administration. The results are discussed in light of the advantages and disadvantages of each administration mode.

Effect of the method of administration, mail or telephone, on the validity and reliability of a respiratory health questionnaire. The Spanish Centers of the European Asthma Study

The European Community Respiratory Health Survey (ECRHS), a multinational survey, assesses and compares the prevalence of asthma among subjects, aged 20 to 44, in several European areas. In Spain, some participating centers have used mail and telephone as methods of questionnaire administration. The objective of the present study was to determine whether the validity and reliability of the questionnaire differed by method of administration. Reliability of the questionnaire was measured with the kappa index and the odds ratio of agreement, and validity with the sensitivity and specificity. This study found differences in the reliability of the questionnaires although these differences were more related to the questions themselves than to the method of administration. Among men, but not women, mailed questionnaires were more sensitive and telephone questionnaires more specific. We hypothesize that these differences in validity were due to the self-selection to more severe symptomatic subjects replying earlier and therefore to the mailed questionnaire. Combining different methods of administration was useful as it increased participation and was an adequate procedure to obtain information of good quality.

A census-based design for the recruitment of a community sample of older adults: efficacy and costs

PURPOSE

This is a report of the scientific and cost implications of a census-based design to identify residents aged 55 and over for a community study of the effects of aging on physical function.

METHODS

A census of residents in a study community was conducted by the use of a mailed questionnaire. For households that did not complete and return the mailed census questionnaire, contact was attempted first by telephone and then by home visit. A comparison was made of the unit costs and characteristics of subjects identified by the different methods.

RESULTS

A total of 3509 age-eligible subjects were identified (78.3% by mailer, 19.5% by telephone, and 2.0% by home visit). Costs per enrolled age-eligible subject were lower for mailing and telephone ($7.76 and $4.72 respectively) than for home visit ($36.25). Subjects identified by home visit were significantly younger than subjects identified either by mail or telephone. After adjustment for age, subjects identified by telephone had less education and income and poorer health and functional status than subjects identified by mail. With the exception of age, there were no significant differences between subjects identified by mailer and home visit.

CONCLUSIONS

A mailed questionnaire with telephone recontact is a practical strategy for community-based recruitment. Recontact of subjects by telephone can be expected to identify subjects who are not well-represented in a sample based only on a mailer. In contrast, the home visit is expensive and identifies subjects who do not differ meaningfully from those identified by mailer.

Smoking in adult survivors of childhood acute lymphoblastic leukemia

BACKGROUND

Health-related behaviors are of particular concern in survivors of childhood cancer as they are at increased risk for second cancers and long-term organ dysfunction. The purpose of this study was to compare the smoking behavior and associated factors in young-adult survivors of childhood acute lymphoblastic leukemia (ALL) with those in sibling controls.

METHODS

A telephone interview that surveyed smoking behavior was conducted with 592 young-adult survivors, treated before age 20 years on Children's Cancer Group ALL protocols, and 409 sibling controls. Using stratified chi-squared analyses and Cox proportional hazards models, we compared the rates of smoking initiation and smoking cessation between survivors and control subjects. Demographic characteristics (age, sex, race, and education) and psychological factors (mood and self-concept) were examined as predictors interacting with survivorship in logistic regression analyses to try to distinguish a subgroup of survivors who may be at greater risk for smoking.

RESULTS

Survivors were significantly less likely to have ever smoked (23.0% versus 35.7%; P<.0001) and thus were less likely to ever be regular, daily smokers than sibling controls (19.1% versus 31.3%; P<.0001). Survivors were less likely to quit smoking than sibling controls (26.6% versus 35.2%), although this result was not statistically significant. There were no interactions between survivor status and either demographic or psychological features on smoking behavior.

CONCLUSIONS

Young-adult survivors of childhood ALL are less likely to experiment with smoking but, once having started, are at similar risk for becoming habitual, persistent smokers as sibling controls.

Telephone vs face-to-face interviews for quantitative food frequency assessment

OBJECTIVE

To develop and test a quantitative food frequency method for administration by telephone.

DESIGN

A comparison study of telephone and face-to-face interviews was conducted among a representative sample of the five major ethnic groups in Oahu, Hawaii. Two interviews were administered 4 to 6 months apart by trained interviewers using identical questionnaires and color photographs of food items showing three different portion sizes. The order of the interviewing methods was randomly assigned. The questionnaire included 115 food items selected to estimate 80% or more of usual dietary intakes. Frequencies and quantities of each item consumed during the past year were obtained.

SUBJECTS/SETTING

Subjects were recruited from the Health Surveillance Program of the Hawaii State Department of Health and consisted of 167 men and 158 women, aged 45 to 74 years, who provided a telephone number. Eighty percent of the face-to-face interviews were conducted in the subjects' homes and 20% were conducted at the workplace or the University of Hawaii Cancer Research Center, if requested.

STATISTICAL ANALYSES

The paired t test was used to compare the mean daily intakes obtained by the telephone and face-to-face methods. Agreement was measured by the intraclass correlation coefficient (ICC), Pearson correlation coefficient and weighted kappa statistic.

RESULTS

The means of energy and each nutrient were slightly higher in the first interview than the second, regardless of the interviewing method. Because of close correspondence among all 3 statistical measures of agreement, only the ICCs are reported. The ICCs ranged from .61 for protein and vitamin A to .69 for dietary cholesterol among men, and from .61 for vitamin C to .74 for saturated fat among women. Agreement was not significantly affected by age, gender, ethnicity, order of interview, or educational level.

APPLICATIONS

Telephone interviews to obtain quantitative food frequencies are cost-efficient methods for estimating usual dietary intakes among persons in widely scattered geographic areas. Photographs of the foods in 3 portion sizes mailed in advance help the respondents estimate amounts eaten.

Caregiving: a common or uncommon experience?

The analysis reported here aims to establish the household prevalence of caregiving in Australia, drawing on a large scale, longitudinal survey conducted as part of the Victorian Carers Project. Comparisons are made with a national survey conducted by the Australian Bureau of Statistics and with Canadian and U.K. findings. Three aspects of caregiving are investigated: reported household prevalence, taking account of differences in definitions used in various surveys; the extent of intergenerational exchanges involved in caregiving; and the time dimensions of caregiving, in terms of duration and patterns of cessation of caregiving over time. A high degree of consistency is found in prevalences of caregiving and implications for the development of policies and programs to support caregivers are raised concerning levels of caregiving, approaches to identifying carers, targeting of services and promotion of caregiving, and the spread of the experience of caregiving across the lifecycle and between generations.

Distributions of smokers by stage: international comparison and association with smoking prevalence

OBJECTIVES

The goals of this study were to describe the distribution of smokers by stage of change in Geneva, Switzerland; to compare this result with distributions observed in other countries; and to assess whether, across samples from different countries, the stage distribution of current smokers was associated with the prevalence of smoking.

METHODS

Two mailed surveys were conducted in Geneva in 1995-1996, in a representative sample of residents (n = 742) and in a representative sample of university members (n = 2,270). A literature review produced seven studies describing the stage distribution in representative samples.

RESULTS

In the Geneva population, 74% of smokers were in the precontemplation stage, 22% in contemplation, and 4% in preparation. In the university sample, the corresponding figures were 72, 20, and 8%. Our results were similar to other European samples, but less favorable than in American samples, where these distributions were typically 40, 40, and 20%, respectively. Across all samples, a low prevalence of smoking was associated with a more favorable stage distribution (r = 0.88, P = 0.002).

CONCLUSIONS

Interventions in Europe need to take into account the large proportion of precontemplators repeatedly observed among smokers. A shift to the right of the distribution of current smokers across stages may help increase quit rates and thereby reduce smoking prevalence. This hypothesis should be tested in prospective intervention studies.

Methods for tracing, contacting, and recruiting a cohort of survivors of childhood cancer

PURPOSE

Due to the use of combined modalities of multiagent chemotherapy, radiation therapy, and surgery, many children with a diagnosis of cancer are now surviving into adulthood. This pilot study sought to determine the feasibility of establishing a cohort of childhood cancer survivors and then to develop methods to trace and contact eligible participants.

MATERIALS AND METHODS

A retrospective cohort design was used. Four hundred and forty subjects who were treated for cancer at the University of Minnesota Hospital before the age of 21, between 1970 and 1986, had survived 5 years, and were alive at last contact were eligible. Tracing efforts were undertaken if the address was more than 2 years old or if a letter was returned by the post office. Contact procedures in this study were designed to determine whether participation rates differed according to the method of contact.

RESULTS

In this cohort of 440 individuals, 11 had died and were not traced. Of the remaining 429 eligible individuals, 408 (95.1%) were successfully contacted. Successful tracing efforts differed by both current age and age at diagnosis. Once contacted, 370 (90.6%) agreed to participate in this study and returned a baseline health questionnaire. Each method of participation, and the combination of methods, showed similar percentages of participation.

CONCLUSIONS

Results from this pilot study show that appropriate methods exist to establish a cohort of adults who have not been contacted since childhood.

Face-to-face household interviews versus telephone interviews for health surveys

The purpose of this study was to compare response distributions in health surveys for two interview modes: face-to-face household interviews and telephone interviews. There were two samples of the Perth metropolitan general population aged 16 to 69 years: a face-to-face household sample (n = 1000) and a telephone sample (n = 222). The samples were generated by probability-based methods commonly used by commercial market research organisations. The surveys occurred in August-September 1992 as part of a larger statewide survey component of a three-year evaluation of the Western Australian Health Promotion Foundation. Respondents were drawn from a two-stage cluster sample based on private dwellings for personal interviews, and from randomly selected listed and unlisted private numbers for telephone interviews. Although the samples did not differ significantly on a number of variables, the telephone sample was significantly higher in residential social status; there was significantly lower reporting of smoking and lower unsafe alcohol consumption in the telephone sample: significantly higher proportions of the telephone sample were in Prochaska's 'action' stage of change for several health behaviours; and there was significantly greater recall of health messages in the telephone sample. Health researchers should treat comparisons between different survey modes with caution, and should be aware that campaign evaluations using telephone surveys and household surveys may yield substantially different results.

The feasibility of a street-intercept survey method in an African-American community

OBJECTIVES

This study evaluates the feasibility of a nonquota, street-intercept survey method that utilized random selection of interview sites.

METHODS

The street-intercept survey was compared with a random digit-dial telephone survey conducted in the same catchment area among African-American adults aged 18 or older.

RESULTS

The street-intercept survey's response rate was 80.2%; residence rate, 85.3%; interview completion rate, 97.9%; interference rate, 4.0%; and yield rate, 2.5 interviews per interviewer per hour. The street-intercept method produced more representative distributions of age and sex than the random-digit-dial survey.

CONCLUSIONS

The street-intercept method is a feasible alternative to traditional population survey methods and may provide better access to harder-to-reach segments of the urban population in a safe manner.

Selection bias from sampling frames: telephone directory and electoral roll compared with door-to-door population census: results from the Blue Mountains Eye Study

Many Australian public health research studies use the telephone directory or the electoral roll as a sampling frame from which to draw study subjects. The sociodemographic, disease-state and risk-factor characteristics of subjects who could be recruited using only the telephone directory or only the electoral roll sampling frames were compared with the characteristics of subjects who would have been missed using only these sampling frames, respectively. In the first phase of the Blue Mountains Eye Study we interviewed and examined 2557 people aged 49 and over living in a defined postcode area, recruited from a door-to-door census. This represented a participation rate of 80.9 per cent and a response rate of 87.9 per cent. The telephone directory was searched for each subject's telephone number and the electoral roll was searched for each subject. Subject characteristics for those who were present in each of these sampling frames were compared with the characteristics of those subjects not included in the sampling frames. The telephone directory listed 2102 (82.2 per cent) of the subjects, and 115 (4.5 per cent) had no telephone connected. The electoral roll contained 2156 (84.3 per cent) of the subjects, and 141 subjects (5.5 per cent) could not be found in either the electoral roll or the telephone directory. Younger subjects, subjects who did not own their own homes and subjects born outside of Australia were significantly less likely to be included in either of these sampling frames. The telephone directory was also more likely to exclude subjects with higher occupational prestige, while the electoral roll was more likely to exclude unmarried persons and males. Researchers using the telephone directory and electoral roll to select subjects for study should be aware of the potential selection bias these sampling frames incur and need to take care when generalising their findings to the wider community.

The effect of nonresponse on prevalence estimates for a referent population: insights from a population-based cohort study. Atherosclerosis Risk in Communities (ARIC) Study Investigators

Characterization of nonrespondents, with the aim of detecting nonresponse bias, is a crucial component of prospective studies. This study was undertaken to investigate the demographic and health characteristics of nonrespondents to a population-based cohort study of cardiovascular disease, to determine whether early-stage nonrespondents differ from late-stage nonrespondents, and to estimate the bias in prevalence estimates for the source population. Sixty-seven percent of eligible subjects completed all phases of the cohort recruitment. Compared to respondents, nonrespondents were less likely to be married, less likely to be employed, and less likely to be well educated. Nonrespondents tended to describe their general health in less favorable terms and were more likely to be smokers. Their reported disease profile, however, was not dissimilar to that of respondents. For several demographic and health characteristics, including marital status, education, and smoking, early-stage nonrespondents differed from respondents more than did late-stage nonrespondents. For example, 42% of early nonrespondents were smokers compared to 37% of late nonrespondents and 22% of respondents. Overall, the bias in prevalence estimates related to nonresponse was small (< 5%) for most of the measured characteristics. Although nonresponse to health surveys is associated with typical attributes, early nonrespondents differ from respondents more than do late-stage nonrespondents. With few exceptions, however, a 33% nonresponse rate did not appear to introduce substantial bias into prevalence estimates for the source community.

Cigarette smoking habits among schoolchildren

STUDY OBJECTIVE

Cigarette smoking is a major preventable cause of morbidity and mortality worldwide. Most adult smokers start smoking regularly some time before 18 years of age. The aim of this study was to determine the age at which children begin cigarette smoking, to study the environmental factors that influence children to smoke, and to understand the reasons why children smoke. The results of this study may help lead to the development of more effective smoking prevention programs.

METHODS

We carried out a cross-sectional survey of all students in grades 6 to 11 (ages: 11 to 17 years) in two high schools in the Jerusalem area, using an anonymous self-completion questionnaire. The students were asked questions regarding the age at which they began smoking, initiation, their smoking habits, their reasons for smoking, and their views on children who smoke. In addition, they were asked about the smoking status of their parents, siblings, and friends. Finally they were asked about the health hazards of smoking.

RESULTS

Of the 847 students who answered the questionnaire, 35% stated that they had smoked at least once and 14% stated that they were currently smoking. The percentage of students who were currently smoking increased gradually with age to 36%. There was a sharp increase in experimental smoking after seventh grade (ages 12 to 13 years). Having a friend who smoked substantially increased the likelihood of smoking, whereas parental smoking or having a sibling who smoked did not increase the likelihood of smoking. The most common reason for starting to smoke was "to try something new" (55%). There was a significant difference between the views of students with different smoking statuses regarding children who smoke: nonsmoking children associated more negative characteristics to smoking. All of the children studied were well aware of the health hazards of cigarette smoking.

CONCLUSIONS

Smoking is highly prevalent among schoolchildren in Jerusalem. The increase in the rate of smoking at the age of 12 years indicates that smoking prevention programs need to be started at an earlier age. Peer pressure is a very strong stimulus for smoking in children. Since children are aware of the health hazards of smoking, prevention policy should be aimed more at associating smoking with negative images and addressing peer pressure.

Mammography use among sociodemographically diverse women: the accuracy of self-report

OBJECTIVE

This study sought to determine the accuracy of self-report of mammography experience among 392 ethnically diverse women aged 50 to 74.

METHODS

Subjects were randomized to the telephone or mail condition and surveyed.

RESULTS

Thirty-one percent of women reported accurately the exact month and year of their most recent mammogram; 54% reported accurately within +/- 3 months, and 83% reported accurately within the year. Greater accuracy was associated with exam recency, White race, and non-Hispanic ethnicity, but not with age, education, or income. Most women could correctly report the reason for, the findings of, and the payor of their mammograms but knew little about how much they or their insurance paid.

CONCLUSIONS

For population surveillance of mammography in the past year, self-report data are generally valid. However, clinical studies requiring more precise dates must use such data with caution. The telephone method, as compared with mail, appears to be a better option for some variables.

Has musculoskeletal pain become less prevalent?

The changes in the one-year prevalence rates of low-back, neck-shoulder and joint pains were studied in Finnish farmers during 1979-1992. A health survey was carried out among a representative sample of Finnish farmers (n = 11,368) with a postal questionnaire in 1979 and in another representative sample (n = 3237) using a computer-assisted telephone interview in 1992. Both surveys inquired for musculoskeletal symptoms during the previous year. The one-year prevalence of low-back and joint pain had decreased in all sex and age groups. The total prevalence of neck-shoulder pain had decreased slightly, but not significantly in either sex. The changes in the occurrence of musculoskeletal complaints from 1979 to 1992 may be attributed to the farming policy of the 1980s in Finland. This study is in line with earlier studies showing that musculoskeletal morbidity per se has not increased in the past decade. Even the reverse seems true, if morbidity is defined as trends in symptom occurrence.

Diversity in the rural poor: differences between households with and without telephones

Differences between households with and without phones in the United States as a whole are well documented, but these differences, and their implications for nursing practice and research, have received little attention in nursing publications. This article 1) reviews findings from national studies of these differences and 2) reports on a nursing study that examined such differences specifically in a random sample (N = 2,053) of low-income families having children eligible for but not using the well-child services of the Medicaid program in rural North Carolina. The study was part of a randomized trial of nursing interventions to encourage parents to use these services. The analyses reported herein focus on how families with and without phones differed in health-related characteristics and in responses to the interventions. The findings have relevance for public health nurses conducting outreach or research with similar low-income families, even when the outreach or research methods do not involve phone contact.

Comparison of clinical and self reported diagnosis for rheumatology outpatients

OBJECTIVE

To examine the sensitivity of patient self reported diagnoses compared with physician diagnoses in a rheumatology outpatient population.

METHODS

A mailed survey to 472 rheumatology outpatients (81% response rate) asked about joint symptoms, disabilities, and underlying rheumatic conditions. The self-reported diagnoses were linked with physician diagnoses in the rheumatology clinic computer based diagnostic registry.

RESULT

Overall there was an 87% sensitivity for self reported compared with physician diagnoses when the matching criteria included compatible yet different diagnoses such as rheumatoid arthritis (RA) and osteoarthritis (OA). The sensitivity for exact match was 65%, and it varied with the underlying clinical diagnosis, and was greatest for RA (90%) and ankylosing spondylitis (AS) (100%), and intermediate for OA (52%) and psoriatic arthritis (50%). The sensitivity of self report was primarily related to the type of diagnosis (RA or AS v other rheumatic conditions; odds ratio = 16.3, 95% confidence interval (CI) 9.0 to 29.5), and also to difficulty in activities of daily living (odds ratio = 2.3, 95% CI 1.1 to 4.6) but not age, gender, duration of disease, or clinic attendance, as shown by multivariate analysis.

CONCLUSIONS

This study in a rheumatology outpatient population indicated that most patients report a diagnosis which is compatible with the clinical diagnosis. These findings give an upper limit to the sensitivity of self reported diagnoses, though further research is needed to assess the extent to which our results may be generalised to other settings.

A twin study of generalized anxiety disorder and major depression

Previous analyses with a sample of female twins sampled from the general population in Virginia have suggested that generalized anxiety disorder (GAD) and major depression (MD) share their genetic determinants but have partly different environmental determinants. The goal of this report is to examine whether these findings apply to samples that include male as well as female twins and contain high proportions of subjects who had been hospitalized for MD. The subjects were ascertained through two different sources: (i) index probands were ascertained through the Swedish Psychiatric Twin Registry for a diagnosis of unipolar or bipolar affective illness; (ii) control twin probands were ascertained through the Swedish Twin Registry. Subjects were sent questionnaires for the assessment of lifetime history of GAD and MD. Positing multinormal distribution of the liability for GAD and MD, we fitted bivariate models to examine the sources of comorbidity. The full model included additive genetic effects, shared environment and individual-specific environment, as well as scalar and non-scalar sex limitations and different thresholds across genders. The best-fitting model included: (i) a genetic correlation of unity; (ii) no common environment; (iii) an individual-specific environmental correlation of 0.28; (iv) different thresholds across genders, but neither scalar nor non-scalar sex-limitations. A model that included additive and dominant genetic effects and individual-specific environment, with correlation of unity for both additive and dominant genetic effects, provided an equivalent fit. These analyses confirm that GAD and MD share the same genetic factors but that their environmental determinants are mostly distinct. Moreover, the present report supports the feasibility of combining clinical ascertained and general-population samples into a single bivariate analysis.

Background factors predicting non-response in a health survey of northern Finnish young men

The study addresses the characteristics potentially predictive of non-response to a health survey among 2500 24-year-old males, on whom a notable amount of other data was available. The overall non-response rate was 40%, part of which was due to the unreachability of some subjects: 4% of the questionnaires were returned because of unknown addresses. Some (n = 39) of the respondents had concealed the identification number in the questionnaire, which made it impossible to link these data to those collected before in their cases. We were therefore able to use effectively the data on 1450 (58%) responding subjects in this paper. Failure to respond was more common among the subjects who had lived in towns in their youth, had not grown up in a complete family, whose socioeconomic status of the family was unknown, and whose mother was young and had a low educational level. Poorer-than-average school performance at elementary school was also predictive of a high non-response rate. Non-response was heavily associated with previous non-response to a health inquiry. Some aspects of health and behaviour in adolescense, such as smoking at the age of 14, were related to non-response to this survey, too. The non-response was higher than average among those subjects who had suffered from mental disorders (serious mental disorders, less serious mental disorders such as neurotic disorders, adjustment reactions, and psychosomatic disorders and mental retardation). The young men who were employed, were students or were doing military service at the age 24 responded better than those who were unemployed or at disability pension.

A prospective cohort study investigating psychosocial predictors of attendance at a mobile breast screening service

This study aimed to examine whether knowledge, attitudes and concerns predicted attendance at the mobile Breast X-Ray Programme in Sydney. A cohort study design was used, whereby women were surveyed prior to the implementation of the program, and two years later records were checked to determine whether they had attended for screening. Telephone interviews were sought with randomly selected women aged 45 to 70 years living in the central Sydney area (the screening van's catchment area). A total of 285 women was surveyed (response rate: 50 per cent). Of these, 86 (30 per cent) subsequently attended at the mobile van and 199 did not. Attendance did not appear to be related to any of the following factors: knowledge; attitudes; prior experience; perceived susceptibility and morbid concern in relation to breast cancer; the amount of information about screening mammography to which a woman had been exposed. The results are interpreted in light of methodological considerations plus findings from our other research.

Survey response rates: national and regional differences in a European multicentre study of vertebral osteoporosis

STUDY OBJECTIVE

This analysis aimed to compare the response rates of those invited to attend for screening in a multicentre, multinational study within Europe.

DESIGN

This was a population survey.

SETTING

Thirty four centres in 16 European countries.

SUBJECTS

Men and women aged 50 years and over were recruited from population based sampling frames to participate in a prevalence survey of osteoporosis. Subjects were invited by post to attend for radiological screening and interview, and non-responders were followed up by repeat mailing.

RESULTS

There was a substantial variation between centres in response rates: the mean was 49% and the range 5-83%. Adjusting for those known to have died or moved house did not affect the overall ranking. The response rates to each mailing also varied between centres: first mailing 45% (range 5-83%) and second mailing mean 10% (range 0-23%). The response rates varied in relation to age and sex and were higher in women than men. Rates fell gradually with age in women but rose in men until the age of 65 years. Response rates varied regionally. These were highest in countries from northern Europe and lowest in southern European countries, but there was wide variation both within regions and within countries.

CONCLUSIONS

Multicentre, multinational studies within Europe will probably become increasingly popular. In this study, despite a standardised approach, the range in response rates between centres both within and between countries was substantial. Attempts at cross national standardisation in survey design can have only a limited effect on yielding uniformity in response.

Relative validity of multiple telephone versus face-to-face 24-hour dietary recalls

The relative validity of multiple telephone 24-hour dietary recalls was evaluated in a feasibility study within the framework of a large prospective investigation on the cause of chronic disease in women. Forty-nine women were interviewed four times both face-to-face and by telephone. Comparison of the total number of calories and intake of protein, carbohydrate, total and saturated fats, cholesterol, fiber, sodium, potassium, calcium, vitamin A, and vitamin C as estimated by multiple face-to-face and telephone interviews revealed an acceptable relative validity for the telephone procedure. Analysis of the position variation in the distribution (percent agreement) comparing the two procedures showed that a change in the distribution of none or one quintile occurs in more than 70% of individuals for all nutrients but vitamin C (69.4%), cholesterol (61.2%), and vitamin A (51.4%). Correlation coefficient analysis showed similar results. Adjustment for nutrient densities did not affect the overall results. Multiple 24-hour telephone dietary recalls appear to be a valid alternative to face-to-face interviews in population studies.

Do blacks and whites differ in their use of health care for symptoms of coronary heart disease?

OBJECTIVES

The purpose of this study was to identify the role of race in seeking and receipt of care for symptoms of coronary heart disease.

METHODS

Data on medical care, sociodemographic characteristics, symptoms, risk factors, income, and insurance were collected in a telephone interview for a random sample of 2030 Black and White adults in inner-city Boston. Rates of care-seeking for symptoms, amounts of delay in seeking care, and rates of receipt of care were compared for Blacks and Whites after adjustment for other characteristics.

RESULTS

Before and after adjustment for other factors, Blacks and Whites were equally likely to seek care. Average delay time was shorter for Blacks, particularly Black women. With the exception of a lower rate of referral to cardiologists among Blacks, receipt of care was similar for Blacks and Whites who sought medical attention for symptoms.

CONCLUSIONS

In an urban population of Blacks and Whites who were similar in socioeconomic status and access to medical care, there were few racial differences in coronary heart disease-related care patterns.

A prospective cohort study investigating the explanation of socio-economic inequalities in health in The Netherlands

In this paper, the objectives, design, data-collection procedures and enrollment rates of the Longitudinal Study on Socio-Economic Health Differences (LS-SEHD) are described. This study started in 1991, and is the first large-scale longitudinal study of the explanation of socio-economic inequalities in health in the Netherlands. The LS-SEHD aims at making a quantitative assessment of the contribution of different mechanisms and factors to the explanation of socio-economic inequalities in health. It is based on a research model incorporating both 'selection' and 'causation' mechanisms, and a wide range of specific factors possibly involved in these mechanisms: health-related life-style factors, structural/environmental factors, psychosocial stress-related factors, childhood environment, cultural factors, psychological factors, and health in childhood. The design of the LS-SEHD is that of a prospective cohort study. An aselect sample, stratified by age, degree of urbanization and socio-economic status, for approx. 27,000 persons was drawn from the population registers in a region in the Southeastern part of The Netherlands. The persons in this sample received a postal questionnaire. An aselect subsample of approx. 3500 persons from the respondents to the postal questionnaire was, in addition, approached for an oral interview. The follow-up of these samples will use routinely collected data (mortality by cause of death, hospital admissions by diagnosis, cancer incidence), as well as repeated postal questionnaires and oral interviews. The response rate to the base-line postal questionnaire was 70.1% (n = 18,973), and that to the base-line oral interview was 79.4% (n = 2802). If the LS-SEHD is compared to a number of frequently cited longitudinal studies of socio-economic inequalities in health from the United Kingdom, it appears that the differences with the OPCS Longitudinal Study and the birth cohort studies (such as the National Survey of Health and Development) are huge. The LS-SEHD is more akin to the Whitehall(I)-study and the West of Scotland 20-07 study. For example it has the sample size of the former but the open population and emphasis on social factors of the latter. A comparison of the results of various longitudinal studies of socio-economic inequalities in health is recommended.

Interviewer versus self-administered questionnaires in developing a disease-specific, health-related quality of life instrument for asthma

We wished to determine the extent to which respondents provided the same answers to a health-related quality of life (HRQL) questionnaire in self- and interviewer-administered forms. One hundred and fifty patients with asthma who were symptomatic or required treatment at least once a week, and had airway hyperresponsiveness to methacholine aerosol (PC20 < 8.0 mg/ml) participated. Patients completed a 152-item HRQL questionnaire in both interviewer and self-administered forms, separated by a 2-week interval, the order determined by random allocation. The percentage of items endorsed by the self-administered approach was significantly higher than that of the interviewer-administered approach overall (46.9 vs 35.8%) (p < 0.0001). The difference was consistent across all six domains; the absolute difference in the proportion of items endorsed varied from 8.9 to 12.3%. The intraclass correlations for the proportion of subjects endorsing an item was 0.84. Self- and interviewer-administered questionnaires yield very similar results in discriminating between subjects, but the self-administered version shows systematically greater HRQL impairment.

Telephone health surveys: potential bias from noncompletion

OBJECTIVES

Little is known about the effect of noncompletion on telephone surveys of health issues. This paper identifies a little-studied source of noncompletion, passive refusal, and evaluates its contribution to noncompletion bias along with two other sources: noncooperation and noncontact. Passive refusals include respondents who repeatedly request callbacks and households where interviewers repeatedly encounter an answering machine.

METHODS

Measures of noncompletion (noncooperation, passive refusal, and noncontact), demographic and socioeconomic characteristics, health risk factors, and indicators of health care access and health status were collected through the Orange County Health Surveys on 4893 respondents. The surveys sampled by random-digit dialing and interviewed by computer-assisted telephone.

RESULTS

Passive refusals have a substantial impact on completion rates and bias due to noncompletion. Commonly used definitions for completion rates may underestimate the bias due to noncompletion because they omit passive refusals. After we controlled for demographic and socioeconomic factors, few noncompletion biases appeared on selected health indicators.

CONCLUSIONS

These results suggest improved reporting of completion rates and support a multivariate framework for studying noncompletion in telephone health surveys.

Reported symptom prevalence rates from comparison populations in community-based environmental studies

National, regional, state, and county rates of mortality and morbidity are frequently used as reference values in epidemiologic studies. However, baseline information on self-reported symptom rates is currently unavailable. This information is potentially very useful as outcome measures in environmental and occupational epidemiologic studies in which symptoms may be the most sensitive measure of effects of low-level chemical exposure. Comparison populations (n = 934 adults) from three community-based studies were used to determine 1-y symptom prevalence rates. Symptoms rates varied greatly between one study, in which a self-administered questionnaire was used, and the two studies for which data were collected via a similar questionnaire administered by trained interviewers. Four symptoms were selected for detailed analysis of covariates associated with symptom reporting: (1) skin irritation, (2) eye irritation, (3) sleep disturbance, and (4) fatigue. Logistic regression analysis of symptom reporting, by demographic data, and a risk perception variable suggested that the prevalence of symptoms varied with gender, race, and degree of respondent's environmental concern. The results from two of three surveys provide rates that might be used as reference rates in a community-based survey if resources are not available to permit the study of a comparison population. These rates may also provide a quick, initial impression prior to embarking on a full-scale epidemiologic study.