Secondary Logo

Journal Logo

Original Research

Reliability and Validity of Self-Reported Symptoms for Predicting Vulvodynia

Reed, Barbara D. MD, MSPH1; Haefner, Hope K. MD2; Harlow, Siobán D. PhD3; Gorenflo, Daniel W. PhD4; Sen, Ananda PhD5

Author Information
doi: 10.1097/01.AOG.0000237102.70485.5d
  • Free

As many as 14 million women in the United States have chronic vulvar pain suggestive of vulvodynia (either generalized or localized, provokable or spontaneous),1–3 and millions of health care dollars are spent annually on this routinely misdiagnosed and inadequately treated syndrome. Only a small proportion of the women with vulvodynia have received the correct diagnosis, and hence, the majority endures ongoing pain and ineffective treatment. Little information exists regarding the clinical history of, risk factors for, or what specific historical, exposure, or pain characteristics might best predict the presence of vulvodynia. Previous perceptions based on clinical experience suggested that most women with vulvodynia would likely be white and would complain of pain with intercourse. However, our data from an Internet-based survey,1 as well as that from a community-based survey,2 indicate vulvar pain consistent with vulvodynia is seen in a similar percentage of African-American and white women. In addition, the history of pain with intercourse is reported in approximately half of all women and does not adequately predict those with ongoing vulvar pain.1 These misconceptions occur because the majority of studies of vulvodynia are being conducted in clinics specializing in vulvodynia, in which only a small subset of women with this disorder are treated. Large epidemiologic studies will allow these and other erroneous assumptions to be revealed and will facilitate identification of other risk factors. However, the diagnosis of vulvodynia, based on survey responses alone, has not been verified. To ascertain the accuracy of predicting vulvodynia or asymptomatic control status among women responding to Internet-based and mailed surveys, we performed a survey of women in the southeastern region of Michigan and invited a subset to come in for a clinical examination to confirm the survey-based diagnosis.


A cross-sectional survey of women in the Southeast region of Michigan, followed by a physical examination for a subset of these women, was performed. Approval for this study was obtained from the University of Michigan Medical School Institutional Review Board. We obtained names, addresses, and, when available, e-mail addresses of 1,447 women members of the University of Michigan’s Women’s Health Registry who lived in a five-county area in Southeast Michigan. This registry consists of women, aged 18 years or more, who have been recruited at meetings, health fairs, etc., in the southeastern region of Michigan, for possible involvement in research. This group, while not necessarily representative of the female population in general, is nonselected for gynecological or vulvar complaints. An abbreviated 67-question Internet-based survey, modified from the 93-item survey used in our previous study,1 was prepared using an online survey tool ( Only those with dyspareunia and vulvar pain completed all 67 questions, with the majority completing only 37 items, due to the skip logic used. Survey questions addressed demographic information, lifestyle factors, medical history, history of pain with intercourse in general, as well as vulvar pain specifically, and presence of other gynecologic symptoms. The online survey was piloted by eight control women and six women with vulvodynia and modified based on their feedback.

The survey was available both in an Internet-based and a written format. The Internet-based survey allowed a skip format that resulted in participants automatically skipping questions that did not apply to them based on their previous answers. It also allowed data to be coded/recorded immediately upon input from the participant. Otherwise, the Internet-based survey and the written survey were identical. Each respondent was asked at the end of the survey whether she would be willing to be contacted for a possible follow-up examination, and if so, to provide contact information.

Participants who received the e-mail invitation were directed via a link to the interactive research web site where the survey was introduced. A short paragraph introduced the study, indicated whom to contact if there were questions or concerns about the survey, and indicated that completion of the survey implied consent.

Those who did not supply an e-mail address were sent a written copy of the invitation and survey (with an enclosed, postage-paid, return envelope), as were those whose e-mail addresses were inaccurate or who failed to respond in any way to the e-mail invitation.

Symptoms associated with vulvodynia include pain at the introitus, often provoked by intercourse or tampon insertion, and lingering pain after intercourse. However, variability in symptom presentation and course among women with vulvodynia is common. Adopted in 2003, the International Society for Vulvar Diseases terminology suggests vulvodynia be divided into localized or generalized vulvar symptoms, and within these, provoked and unprovoked subsets.4 Based on data suggesting considerable overlap in these characteristics,5,6 we included all women with vulvodynia, regardless of pain being strictly localized or being more generalized, or whether symptoms were provoked or unprovoked. We made the presumptive diagnosis of vulvodynia based on characteristics included in the description of vulvodynia, including the report of pain at the introitus—described as “at the opening of the vagina”—that has lasted a minimum of 3 months. This descriptive localizing phrase was used to identify ongoing pain in the most common area of discomfort for the woman without using the phrases of “inside” and “outside” which, in our experience, correlate poorly with the anatomical location of the sensitivity. Controls were defined as women reporting no current pain with intercourse and no history of vulvar pain lasting 3 months or longer.

We used strict definitions for the diagnoses of vulvodynia and for asymptomatic control status to maximize the accuracy of predicting cases and controls, as is needed for clinical and epidemiological studies. A substantial proportion of the women responding to the survey would not be classified as having vulvodynia or as being asymptomatic controls. Examples include women with vulvar pain lasting less than 3 months, women with current pain with intercourse in another location, etc. These women may, with further evaluation, be found to have vulvodynia and, hence, could not be categorized as a control, but they did not meet the criteria for a case. Hence, they were not included in the comparison of women predicted to have vulvodynia and those predicted to be controls.

A subset of women predicted by survey responses to be cases or to be controls, who had indicated a willingness to be contacted by the investigators, were called by the research assistant and invited to participate in the examination portion of the study. Upon arrival at the office of one of the investigators (B.D.R.), each woman again had the study explained, read the informed consent, had any questions answered, and signed the consent for the examination.

At this visit, the woman’s history of pain was reviewed, and an abdominal and pelvic examination was performed. The vulvar area was examined for any dermatologic changes such as color changes (whitening, erythema, abnormal pigmentation), lesions, or swelling. Sensory testing at 27 sites of the perineum/introitus, using the dacron swab pressure test, was conducted to determine sensitivity of the vulvar region (labia majora, interlabial sulci, introitus) and hymenal remnants. At each site, the woman indicated if she had discomfort and, if so, the intensity (mild [1], moderate [2], or severe [3]) and the character of the discomfort. A speculum examination was then performed, and vaginal secretions were obtained for in-office laboratory and Candida culture. The vaginal secretions were evaluated for 1) pH; 2) normal saline prep assessing epithelial cells, white blood cells, background flora, presence of clue cells or Trichomonas; 3) amine odor after application of potassium hydroxide; and 4) potassium hydroxide preparation looking for budding yeast or hyphal forms.

Vulvodynia, diagnosed with the inclusion of examination data, was confirmed if the physical examination indicated: (1) no evidence of a dermatological disorder (other than erythema), (2) no evidence of an infectious disorder (clinically evident herpes simplex virus, condyloma, or acute Candida vulvovaginitis), and (3) a positive finding of dacron swab tenderness at the introitus or hymenal remnants or both (defined as a score of 3 or more among the introital and hymenal sites tested—which might be reached by a minimum of one site with severe discomfort, or three sites with mild discomfort). Although an occasional woman with generalized, spontaneous vulvodynia will have a negative dacron swab test, the majority will not, and hence a positive finding is very helpful in diagnosing even this category of vulvodynia. If there was evidence of an active Candida infection, the history was reviewed to determine whether eradication of active yeast infections in the past had occurred, and whether this was associated with symptom resolution. If treatment had occurred in the past, and symptoms were unchanged, the current presence of Candida did not refute the diagnosis of vulvodynia.

If the woman had vulvodynia confirmed by the office history and examination, we answered any questions, and provided references from which the participant could learn further about this entity. If she desired further diagnosis and treatment, we referred her back to her health care provider. If this provider wished us to further evaluate and treat the patient, she was scheduled at one of our vulvar clinics for routine evaluation and treatment.

Data entered online by participants were downloaded and were placed in an SPSS (SPSS Inc., Chicago, IL) data file. Data from the physical examination and from the written surveys were entered into the SPSS database. Identification of survey method (Internet-based or written) was included in the file to allow comparison of data from each group.

The response rates to the surveys were calculated, and the frequencies of characteristics of the participants were determined.

The diagnosis (by survey) of vulvar pain suggesting vulvodynia, and that of controls, were made using the criteria listed above. Independent samples t test or Pearson χ2 tests were used to compare continuous or categorical characteristics between the two diagnosis groups.

The reliability of the history of vulvar pain as reported on the survey compared with that obtained at the examination was determined using the Cohen’s κ statistic. The validity of the diagnosis of vulvodynia as predicted by the survey responses (suggested vulvodynia) was estimated by the examination-based diagnosis, again using the Cohen’s κ statistic. The sensitivity and specificity were calculated and the overall accuracy determined.


Between August 5, 2004, and December 13, 2004, 1,181 of 1,447 women in the Women’s Health Registry who had provided an e-mail address were sent online surveys, 82 (6.9%) of the surveys were returned as undeliverable, and 614 (51.9%) were completed and returned, for a 55.9% response rate. The nonrespondents and those undeliverable (n=567), as well as those who had not provided an e-mail address (n=266) were sent written surveys, and a written reminder 2 weeks later if their survey had not been returned. Forty-nine (5.9%) of the written surveys were not deliverable. The response rate for the written survey was 55.1% (432), with an overall response rate for the study (Internet-based and written combined) of 72.3%. The response rates for the Internet-based and the written surveys were not significantly different (55.9% and 55.1% response, respectively, P=.74).

Respondents ranged from 19 to 92 years of age, with the mean and median age of 45 years (Table 1). A small majority had had at least 4 years of college (66.1%), and 59.3% were married. A total of 899 (85.9%) were white, 95 (9.1%) were African American, and 52 (5.0%) were of other ethnicities. Of the 1,046 women, 96.9% had been sexually active. The women with vulvodynia predicted by the survey were younger, more likely to be married, and tended to rate their own health slightly lower than women without vulvar pain. They were, however, similar in educational background and ethnicity to those without pain.

Table 1
Table 1:
Characteristics of All Respondents and Vulvodynia Cases and Controls Predicted by the Survey

Of the 1,046 participants, 621 (59.4%) women reported having had pain with intercourse in the past, and 233 (22.3%) indicated that this was still a problem for them (Fig. 1). Four hundred one (38.3%) of the women reported a history of vulvar pain (pain at the opening to the vagina), with 107 (10.2%) reporting that this pain had lasted more than 3 months. Among the 107 with prolonged vulvar pain, 28 (26.2%) thought that their pain was now resolved, with 79 (73.8%) reporting the vulvar pain was still present. The latter group of 79 women was given the presumptive diagnosis of vulvodynia based on the survey. Those given the presumptive diagnosis of being a control included 543 (51.9%) women who denied both current pain with intercourse and ongoing pain at the introitus.

Fig. 1.
Fig. 1.:
Various types of genital pain reported (N=1,046).Reed. Survey-Based Diagnosis of Vulvodynia. Obstet Gynecol 2006.

The remaining 424 women (40.5%) did not meet the criteria for either cases or controls. Of these, complete data on which to predict their case/control status (data on ongoing dyspareunia or whether they had had vulvar pain) were available on 390 women. These women could not be classified as controls because 171 (43.8%) reported ongoing pain with intercourse, and an additional 219 (56.2%) reported vulvar pain (either not lasting for 3 months, or currently resolved). These 390 women also did not meet criteria for vulvodynia. Seventy-six (19.5%) reported pain with intercourse, but not located at the opening to the vagina. An additional 83 (21.3%) reported pain with intercourse and vulvar pain, but denied duration of 3 months or more. Twelve (3.1%) reported vulvar pain lasting at least 3 months, but reported that it had resolved, and 219 (56.2%) reported no pain with intercourse at this time.

Overall, 85.7% of those classified as having vulvodynia, and 66.0% of those classified as asymptomatic controls, indicated they were willing to be contacted to potentially come in for an examination. From the first 40 patients classified as having vulvodynia and the first 60 classified as controls, 39 women classified as having vulvodynia and 57 classified as controls were reached by telephone and were invited to come in for an examination. Twenty-eight women (71.8%) classified as having vulvodynia and 34 pain-free controls (59.6%) were interviewed and examined in the office setting (P=.22). None were found to have a dermatologic diagnosis (such as lichen sclerosis). Three (4.8%) had a positive culture for yeast—one woman with vulvodynia and two controls. Of these, none reported itching or discharge in the past 3 months.

The history taken at the office demonstrated very good agreement with that reported on the survey. Of the 28 women predicted to have vulvodynia based on the survey, who were also seen in the office to verify the diagnosis, 26 (92.9%) reported current pain at the opening to the vagina that had lasted at least 3 months when interviewed in person, indicating the survey-based history had excellent reliability (Cohen’s κ statistic=0.86, 95% confidence interval [CI] .73–.99).7 Of those reporting on the survey to not have current pain with intercourse and to never have had vulvar pain lasting at least 3 months, 32 (94.1%) reported no pain with intercourse and no history of ongoing vulvar pain when questioned at the examination.

The agreement between the presumptive diagnoses based on the survey classification and the office diagnosis (history and physical examination) is shown in Table 2, part A. The overall accuracy was 88.7%, with a Cohen’s κ of 0.78 (95% CI 0.64–0.92). Both the sensitivity and the specificity of the survey classification were high. In addition, 27 of 28 (96.4%, 95% CI 81.7–99.9%) of those who were classified as having vulvodynia by the survey were confirmed at the office visit, while 28 of 34 (82.4%, 95% CI 65.5–93.2%) women classified as controls were asymptomatic at the examination.

Table 2
Table 2:
Accuracy of Different Diagnostic Criteria Compared With Clinical Examination

Review of the survey results indicated that, although many women denied that their vulvar pain had lasted 3 months or more, often they then reported their symptoms had not “resolved” and rated their pain at the time of the survey and in the past 6 months as more than 0. We, therefore, assessed the impact of combining those who reported “ongoing pain” with those who reported “more than 3 months of pain.” A total of 350 of 893 women with complete data (39.2%) fit this expanded criteria, and a total of 56 of these women were examined in the office. Agreement between the predicted diagnoses using this expanded criterion was 84.4% (Cohen’s κ=0.673, 95% CI 0.51–0.83) (Table 2), with slightly improved sensitivity (88.9% sensitivity) but substantially decreased specificity (77.8%).

The test characteristics of these two diagnostic criteria for vulvodynia, as well as the criteria for controls, are shown in Table 2. Our original criterion (women who had reported vulvar pain at least 3 months in duration) identified women with vulvodynia with fewer false positives. Similarly, to identify controls with the minimum of false negatives, our original definition of no vulvar pain lasting more than 3 months and no current pain with intercourse was good. If identifying the maximum number of women with vulvodynia were necessary, using the criteria of either reporting pain for more than 3 months, or currently having vulvar pain with intercourse (without the duration requirement) were most sensitive (88.9%), but the specificity decreased with this criteria.

Of the 79 women predicted by the survey to have vulvodynia, 56 (70.9%) had seen a health care professional regarding this pain in the past 2 years, and 44 (55.7%) had been given a diagnosis for this pain. Of these, only 25 (31.6% of the 79 women, and 56.8% of those given a diagnosis) had been given the diagnosis of vulvodynia or vestibulodynia or both. By definition, our strict definition identified all 25 of those who had been given that diagnosis (100.0%). Twenty-nine women who were predicted by our survey to have vulvodynia, and who had seen a health care provider for vulvar pain in the past 2 years were among those examined in our office. Five of these women were among those who were told they had vulvodynia—all five were found to have this diagnosis confirmed at the examination. Of the 24 women who our survey predicted to have vulvodynia, but who were not given this diagnosis when seen by a medical provider in the past 2 years, 23 (95.8%) had vulvodynia confirmed at the study office visit.

Compared with women answering our written survey, women responding to the Internet-based survey tended to be younger (42.0 versus 50.1 years, P<.001), more educated (73.8% versus 59.4% with a 4-year college degree or greater, P<.001), and had a higher household income (53.4% versus 47.5% household income greater than $60,000, P=.05). However, the two groups did not differ in a positive history of pain with intercourse (62.4% versus 58.4%), ongoing pain with intercourse (39.0% versus 31.7%), vulvar pain (37.0% versus 36.6%), vulvar pain lasting more than 3 months (21.3% versus 24.8%), and of those, the proportion who reported pain resolution (27.3% versus 29.0%).


This study defines a survey-based approach for identifying vulvodynia cases and asymptomatic control women and validating these findings by examination. Because many women with symptoms and signs of vulvodynia have not been given this diagnosis, it would be difficult to identify cases for research based solely on asking if they have this diagnosis or by checking diagnosis codes on medical records. This study assessed the reliability and validity of diagnostic criteria determined on a self-administered survey, compared with that confirmed by an office visit. We found an excellent agreement between the prediction based on the survey and that made in the office when identifying a case as a woman reporting pain at the opening to the vagina that has been present for a minimum of 3 months.

Others have addressed the accuracy of the vulvodynia diagnosis. Masheb et al8 recently demonstrated that the use of two criteria—the presence of pain on vaginal entry (indicated by pain on speculum insertion) and tenderness to pressure localized within the vulvar vestibule (indicated by the dacron swab test)—demonstrated good reliability and validity when making the diagnosis of vulvodynia. Excellent physician test-retest reliability was found for ratings of pain on speculum insertion (r=0.63, P<.001) and dacron swab testing (r=−0.57, P<.001), but not for erythema ratings. They also demonstrated that findings on examination correlated with information given on their survey. For example, the pain on speculum insertion was correlated with reported dyspareunia (r=−0.40, P=.004), and the dacron swab test was associated with the total McGill Pain score (r=0.30, P=.03), as well as the sensory (r=0.30, P=.04) and affective (r=0.30, P=.05) subscales of the McGill Pain Questionnaire, suggesting sexual pain and affective pain descriptors are related to the examination findings that were found to be most reliable.

Bergeron et al9 have assessed the accuracy of the diagnosis of vestibulodynia (vulvar vestibular syndrome) by assessing test-retest reliability and inter-rater agreement. They found that the Cohen’s κ values for the vulvar vestibulodynia diagnosis ranged from 0.66 to 0.68 for inter-rater agreement (with correlation of pain rating from dacron swab testing between the gynecologists) and from 0.49 to 0.54 for test-retest reliability. Again, ratings of vulvar erythema did not correlate significantly with respect to either inter-rater agreement or test-retest reliability. They found that of Friedrich’s three diagnostic criteria (physical findings confined to vestibular erythema, pain on touch to the vestibule or with intercourse, and tenderness to pressure at the introitus), only tenderness to pressure within the vulvar vestibule differentiated dyspareunia patients with and without vestibulodynia. They concluded that vulvar vestibulodynia can be reliably diagnosed in women with dyspareunia, and that pain could be rated and described in a consistent fashion by women with this disorder.

Similarly, Munday et al10 demonstrated that erythema of the vestibule was not predictive of vulvodynia or severity of sensitivity to the dacron swab test or to the algesiometer, or to history of pain with intercourse. In addition, they and others11 have suggested that sensitivity at the introitus can be seen among symptom-free women. Whether these women with sensitivity, but no symptomatic complaints, share the pathophysiology associated with vulvodynia, or whether they identify women at risk for symptomatic vulvodynia, is unknown.

This study was designed to assess the accuracy of survey-based data for identifying women with vulvodynia. B. Harlow et al12 have assessed the prevalence of vulvar pain consistent with vulvodynia via surveys. They used the criteria of 3 months or more of one of the following: (1) burning pain in the genital area with or without chronic itching, (2) knifelike or sharp pain in the genital area with or without burning or itching, or (3) excessive pain on contact when inserting tampons, during sexual intercourse, or during pelvic examinations. After an additional telephone interview, 200 women were predicted to have vulvodynia, and of these, 70 underwent a physical examination, of whom 80% were confirmed to have the diagnosis.

We found that the accuracy of the prediction of vulvodynia was very good for those who responded to the Internet-based and written surveys, and that, although women responding to the Internet-based survey differed in age, education, and income from those answering the written survey, they did not differ in characteristics of their pain with intercourse and vulvar pain. Others have compared Internet-based surveys to more conventional surveys. Cronk et al13 found the attrition rate was less when participants were randomly assigned to a take-home Internet-based survey compared with an in-class or take-home written survey, but the answers to the survey questions did not differ significantly. Another study compared Internet-based with written survey participation and responses among college students and found no difference in participation rates. However, those completing the Internet-based survey were more likely to answer socially threatening questions.14,15 We had a similar percentage of women respond to both the Internet-based and written survey.

This study addressed the validity and reliability of the diagnosis among women meeting strict definitions of vulvodynia or control status. Being able to identify substantial numbers of these women for studies will allow epidemiologic assessment that would be expected to substantially improve our understanding of this clinical problem. A large proportion of respondents could not be classified by this survey as having vulvodynia or being a control—typically because the presence of genital pain symptoms that were not of the location, quality, or duration expected in vulvodynia. While some of these women would be expected to have vulvar hypersensitivity, with or without other pathology, further evaluation with a detailed history, and a physical examination would be required to make a clinical diagnosis among these women.

Limitations to this study exist. The sample was obtained from a registry consisting of women volunteering to be approached regarding Women’s Health Issues. They were not selected for genital symptoms, but potential participation bias cannot be excluded. The nature of this group limits the generalizability of the results to similar populations. The numbers of women examined were limited. In addition, a substantial proportion of respondents were not symptom-free and could not be categorized as a control, but did not meet criteria for vulvodynia either, thereby creating a group of women who may have other diagnoses (chronic pelvic pain, endometriosis, etc.) or may actually have vulvodynia. Hence, the prevalence of vulvodynia found in this study is a minimal estimate of that in the population. However, the associations between the predicted and the verified diagnoses for those estimated to have vulvodynia and those predicted to be controls, based on survey responses, were excellent, and support the use of these methodologies for identifying and following up with a large number of women with vulvodynia.

Therefore, we have demonstrated that it is feasible to predict the vulvodynia diagnosis as well as to identify control women with good accuracy, based on survey responses about genital symptoms, using either an Internet-based or written survey methodology. This survey methodology is one that will allow identification of a large number of women with (and without) vulvodynia. Data from these larger groups of women can then be used to further elucidate risk factors associated with this disease, as well as factors associated with remission or response to treatment.


1. Reed BD, Crawford S, Couper M, Cave C, Haefner HK. Pain at the vulvar vestibule: a Web-based survey. J Low Genit Tract Dis 2004;8:48–57.
2. Harlow BL, Wise LA, Stewart EG. Prevalence and predictors of chronic lower genital tract discomfort. Am J Obstet Gynecol 2001;185:545–50.
3. Harlow BL, Stewart EG. A population-based assessment of chronic unexplained vulvar pain: have we underestimated the prevalence of vulvodynia? J Am Med Womens Assoc 2003;58:82–8.
4. Moyal-Barracco M, Lynch P. 2003 ISSVD terminology and classification of vulvodynia: a historical perspective. J Reprod Med 2004;49:772–7.
5. Edwards L. Subsets of vulvodynia: overlapping characteristics. J Reprod Med 2004;49:883–7.
6. Reed BD, Gorenflo DW, Haefner HK. Generalized vulvar dysesthesia vs. vestibulodynia: are these distinct diagnoses? J Reprod Med 2003;48:858–64.
7. Fleiss J, Levin B, Paik M. Statistical methods for rates & properties. 3rd ed. New York (NY): John Wiley and Sons, Inc. 2003.
8. Masheb RM, Lozano C, Richman S, Minkin MJ, Kerns RD. On the reliability and validity of physician ratings for vulvodynia and the discriminant validity of its subtypes. Pain Med 2004;5:349–58.
9. Bergeron S, Binik YM, Khalife S, Pagidas K, Glazer HI. Vulvar vestibulitis syndrome: reliability of diagnosis and evaluation of current diagnostic criteria. Obstet Gynecol 2001;98:45–51.
10. Munday P, Green J, Randall C, Fox D, Singh M, Ismailjee F, et al. Vulval vestibulitis: a common cause of dyspareunia? BJOG 2005;112:500–3.
11. Eva LJ, Reid WM, MacLean AB, Morrison GD. Assessment of response to treatment in vulvar vestibulitis syndrome by means of the vulvar algesiometer. Am J Obstet Gynecol 1999;181:99–102.
12. Harlow BL, Stewart EG. Adult-onset vulvodynia in relation to childhood violence victimization. Am J Epidemiol 2005;161:871–80.
13. Cronk BC, West JL. Personality research on the Internet: a comparison of Web-based and traditional instruments in take-home and in-class settings. Behav Res Methods Instrum Comput 2002;34:177–80.
14. Pealer LN, Weiler RM, Pigg RM Jr, Miller D, Dorman SM. The feasibility of a web-based surveillance system to collect health risk behavior data from college students. Health Educ Behav 2001;28:547–59.
15. McCabe SE, Boyd CJ, Couper MP, Crawford S, D’Arcy H. Mode effects for collecting alcohol and other drug use data: Web and U.S. mail. J Stud Alcohol 2002;63:755–61.
© 2006 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.