Junkfood Science: Survey Says: Parents purportedly want to be screened for alcohol use by their child’s pediatrician

November 20, 2008

Survey Says: Parents purportedly want to be screened for alcohol use by their child’s pediatrician

Pediatricians are being encouraged to add another screening test to their lifestyle assessments of the families of their young patients. Medscape just reported on a study in the current issue of the journal Pediatrics telling pediatricians that it had found eight out of ten parents would “welcome or not mind at all” their child’s doctor asking them about their alcohol use and screening them for a drinking problem. Pediatricians also read that one-third or more parents said they would similarly appreciate their child’s pediatrician reporting them to a social worker or making an appointment for them with their own doctor if he/she found they had a drinking problem.

“The fact that more than 75% of the parents who screened positive were receptive to alcohol screening should reassure pediatricians who fear a negative response to questioning,” the study’s lead author Dr. Celeste R. Wilson, M.D., from the Children's Hospital in Boston, Massachusetts, was quoted saying in Medscape Psychiatry.

Parents, do these results describe you? Do most parents really want their child’s pediatrician questioning them about their own habits when they take their child to the doctor? And does this study provide reassurance for pediatricians hesitant to adopt alcohol screening in their practices?

Going to the actual study, finds it was a survey. Opinion polls are traditional marketing technique, as findings are easily varied depending on how the survey population is selected, how the forced choice questions are asked and interpreted, and what is/isn’t asked or reported. This paper presented a singular viewpoint promoting alcohol screening tests and a closer look reveals why it failed as sound medical research and proved unable to validate alcohol screening by pediatricians. To earn the free continuing medical education credits offered from Medscape, by the way, physicians merely had to answer two questions about which method parents preferred to be screened and alcohol abuse among parents.

No one would deny that children in households with severe substance abuse are at increased risks, and healthcare professionals are already obliged to seek help for any children they believe to be in danger of physical or emotional harm or neglect. That is, of course, very different from adding universal screening for alcohol misuse to preventive care guidelines for pediatricians. With today’s clinical guidelines no longer guidelines for physicians but mandated performance measures and compulsory electronic reporting for third-party monitoring, and the fact that blanket screenings potentially impact millions of families, sound evidence of efficacy and careful, objective analyses for potential harms are imperative for doctors and parents.


In the study introduction, Dr. Wilson and colleagues began by saying that pediatricians were in a unique position to explore family situations and screen parents for the nature and extent of alcohol use, discuss with parents the effects of positive and negative role modeling on their children, and refer them for treatment and follow-up.

For this observational study, they recruited participants from three pediatric clinics: one urban hospital-based clinic in Cambridge, Massachusetts; one suburban private practice in Milton, Massachusetts; and one rural clinic in Burlington, Vermont. From the more than 115,000 childcare visits at these clinics between June 2004 and December 2006, the authors said they invited only 1,130 parents to participate.

No information on how these invites were selected, other than the convenience of the available research assistant, and why the other parents weren’t. There was no randomized or arbitrary selection process and no way to know if these parents were representative of parents being seen at these clinics or of parents across the country. Selection bias, alone, invalidates the findings.

The authors then excluded another 102 parents who didn’t read or understand English, leaving 1,028 eligible parents. Only 929 agreed to participate and be screened for alcohol usage screening and answer questions about alcohol screening. But of those, only 879 parents among the 115,000 childcare visits completed the questionnaires. The 15% of selected parents who then refused to be screened or participate were not included in the author’s evaluation of the receptiveness of alcohol screening.

Lending additional caution in generalizing the readiness to be screened, the effectiveness of the screening tools, and the number of parents admitting alcohol use that were reported in this study, is the fact that these study participants answered the surveys anonymously in the form of mail-in questionnaires. In contrast, the identify of parents in a clinic screening program would be known.

The questionnaire included 40 questions of demographic information, preferred screening methods, preferred interventions should screenings be positive, and the parents’ acceptance of screening, along with two brief self-administered alcohol screening tests: TWEAK and AUDIT.

TWEAK is a five-item questionnaire, in which the authors asked their drinking Tolerance, if others have expressed Worry about their drinking anytime in the past year, if they’ve ever had a drink in the morning as an Eye-opener, if others have ever been Annoyed by their drinking (different from other studies which use reports of Amnesia for the “A”), and if they’ve ever felt the need to Kut down their drinking. They also used a ten-item AUDIT scoring test, which asked about the amount and frequency of drinking, alcohol dependence, and problems caused by alcohol. For this study, the authors used a TWEAK score ≥2 and ≥3 or an AUDIT score ≥6 and ≥8 for women and men, respectively.

Alcohol screening tools — are they valid for pediatric practices?

These screening tests vary in their sensitivity and specificity, depending on the setting (emergency rooms with problem alcoholics to general population) and population (gender, age, race/ethnicity) they’ve been studied on, as well as the scoring point cut-offs used.

According to the medical literature, TWEAK was developed to be a screening tool to evaluate alcohol use among pregnant females. Another recent RWJF study of female VA patients in Seattle, Washington, evaluating TWEAK against DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition) alcohol abuse or dependence diagnostic criteria, concluded: “TWEAK has low sensitivity as an alcohol-screening questionnaire among female VA outpatients and should be evaluated further before being used in other female primary care populations.” For their population, the AUDIT was more suitable in identifying current hazardous alcohol abuse.

Dr. Cheryl Cherpitel, DrPH, Adjunct Professor at the Community Health & Human Development School of Public Health at the University of California, Berkeley, has published numerous articles on research examining the validity of alcohol screening tests conducted for the National Institute on Alcohol Abuse and Alcoholism at the NIH. She’s reported that the validity of screening tools are inconsistent and vary considerably across ethnic and gender groups. All screening tests also have lower sensitivity for women compared to men. The appropriateness of various screening tests appear to necessitate clinical judgment and individualization. This illustrates the dilemma physicians encounter with one-size-fits-all performance measures imposed by third-party payers.

As Dr. Cherpitel reviewed in Alcohol health & Research World, the sensitivity of AUDIT is about 72% for women and 93% for men; and the sensitivity of TWEAK is 74% for women and 94% for men. Sensitivity refers to the percentage of alcoholics who would be correctly identified by a screening test. As sensitivity increases, she said, specificity usually drops, meaning people without alcohol problems would be falsely identified as having a drinking problem. So in the attempt to identify as many problem drinkers as possible, that means significantly more false positives and more people will be wrongly tagged for intervention and follow-up.

The sensitivity of TWEAK has also been shown in research to be lowest in the primary care setting (75%), followed by the general population (83%) and highest when studied among emergency room patients, she has reported. The cut-off values make a difference, too. With a cutoff score of 2, TWEAK was shown having a sensitivity of 79% and a specificity of 83% for identifying pregnant women who consume one or more ounces of absolute alcohol per day. Using a cutoff value of three points, brought the sensitivity up 90% and the specificity down from 75 to 80%.

A recent review of AUDIT’s performance, according to Dr. Cherpitel, found its sensitivity ranged from 38 to 94% and its specificity ranged from 66 to 90%. The performance of these tools vary significantly, she said.

Not desirable qualities for a screening test.

It can be popular to believe that “I don’t have a problem, so there’s nothing to be afraid of by being screened.” But, this belief ignores false positives in screening tests, and these specificities are considerably higher than other screening tests, meaning large numbers of parents and families would be reported and entered into electronic medical records as potentially having drinking problems.

Survey participants

The Boston researchers reported that most of the participants in their poll were women (82.9%), white (813.%), married (70.7%) and college graduates (57%). Even so, unlike other studies of these screening tools, they found no significant variations in the positive alcohol screening results across any demographic characteristic (race, age, gender, socioeconomic status, married status, pregnancy, children’s ages, etc.) or clinic setting.

They reported that 1 in 9 parents (11.5%) had a positive score for having a drinking problem. Dr. Wilson and coauthors stated that they believed that the percentage of drinking problems they identified “is likely an underestimation of the actual prevalence in families.”

Yet, this is a prevalence about 3% higher than among all U.S. adults 18 years and older (8.3%) reported in the latest Health, United States 2007 report issued by the Secretary of the Department of Health and Human Services.

Parents’ acceptance of alcohol screening

According to Dr. Wilson and colleagues, 89% of parents overall said that they would “welcome or not mind at all” being asked about their drinking habits during the course of a pediatric office visit. All of the parents with positive screening tests, however, were less likely to welcome screening than those with negative scores (91% compared to 77%). “This finding should reassure pediatricians, because fear of a negative response to questioning has been cited as a potential barrier to alcohol screening,” they concluded.

Overall, women were reported as being more accepting of alcohol screening and interventions than the men in this study. Among those screening positive for alcohol misuse, reportedly 84% of women and 75% of men welcomed or wouldn’t mind at all if the pediatrician talked to them about it. About 78% of women would welcome being given contact information for treatment centers and educational material, compared to 55% of men. More than half of men (57%) wanted no intervention, compared to 26% of women.

While the authors reported that significant percentages of alcohol-positive parents welcomed their child’s pediatrician notifying their own doctor or making an appointment with their doctor, and reporting them to a social worker, the numbers who said they would be "mildly to extremely annoyed" were larger. This flipside wasn’t pointed out in the paper. Seven out of ten men said they would be annoyed having the pediatrician notify their own doctor or schedule an appointment for interventions, and [only] eight out of ten would be annoyed if the pediatrician reported them to social services. Seven out of ten women would be annoyed at these interventions.

The authors admitted that they didn’t specifically ask why more than one in four parents were not receptive to screening and why most would be upset at being referred to physicians or social services. They suggested that perhaps parents feared losing custody of their children, although they said that because the responses were anonymous in this study, such concerns would likely be more significant for screening outside this study.

Of the various methods for administering the screening tests, the authors said parents most preferred them to be administered by the pediatrician (73% and 52% of women and men, respectively), followed by computer or paper questionnaires. Who administered the screening test made little difference in the parents who reported they would likely be honest, overall about 86%.

The authors concluded that parents with alcohol problems are accepting of alcohol screening in pediatricians’ offices and the preferred method of intervention was having the pediatrician talk to them about their drinking. “If pediatricians can implement a screening and intervention model, then there is already a population of parents who could surely benefit and may, indeed, be willing to take that first step.”


No mention was made if screening leads to effective interventions that reduce drinking problems and result in favorable changes to the home environment for the children. Screening tests without demonstrated beneficial outcomes defeat their purpose.

Pediatricians didn’t learn that the U.S. Preventive Services Task Force found in both its 1996 and 2004 reviews of the evidence “that counseling interventions had mixed results on the long-term health outcomes of adults. No studies found statistically significant, long-term effects on morbidity.” Only 10-19% people were no longer drinking at harmful or risky levels compared to controls after 6-12 months of multiple clinical interventions. Most studies of brief motivational counseling by clinicians in a busy primary care population resulted in minimal effectiveness in reducing alcohol consumption.

The USPSTF found no support for screening or behavioral counseling interventions among adolescents. Some studies of younger populations have even found moderate increases in drinking compared to controls. And “more research into the efficacy of primary care screening and behavioral interventions for alcohol misuse among pregnant women is needed.” While the USPSTF made B recommendations for screening adults in primary care settings, no studies have evaluated benefits versus risks when an adult is screened by a doctor who is not their own care provider.

The USPSTF also “found only two poor-to-fair quality studies evaluating the cost-effectiveness of alcohol behavioral counseling interventions.” Interpreting the medical literature is difficult because of inconsistent definitions and outcome, it stated, although cost savings due to reductions in ER and hospitalizations might be possible. The study in Pediatrics didn’t estimate the additional costs and time for pediatricians to screen and counsel parents and the impact on other needed pediatric services.

Most importantly, there was no discussion by the Boston authors of the potential harms of alcohol screening. The USPSTF and the 2004 systematic evidence review prepared by the Oregon Evidence-based Practice Center for the Agency for Healthcare Research and Quality, reported that no research to date has examined the adverse effects resulting from false positives, or the negative effects of generalized screening for alcohol misuse. Besides highlighting the lack of research on the potential harms of screening, one significant concern they raised is that several studies have reported significantly greater drop-out rates among people who object or are offended by screening. This could be critically important for pediatricians who risk turning away those large numbers of parents who resent the unwanted intrusion and could stop bringing their children in for well-child checks or care.

Once again, a survey is not medical research. It is not a randomized clinical trial designed to be a fair test of the effectiveness of a medical intervention and demonstrate benefits that outweigh risks for harm. Opinion polls don’t make evidence-based clinical practice because consensus of opinions is not science. This survey provided healthcare providers with no support for screening or for its inclusion in clinical practice guidelines for pediatric populations.

© 2008 Sandy Szwarc


This study was funded by Robert Wood Johnson Foundation - Substance Abuse Policy Research Program and Dr. Wilson is supported by a faculty fellowship grant from the Center of Excellence in Minority Health and Health Disparities Faculty Fellowship Program at Harvard Medical School.

Bookmark and Share