To the Editor:
In the July 1998 issue of the Journal, we reported findings from the Upper Peninsula Research Network’s (UPRNet) height and weight study. In addition to the data we reported, we wondered whether the reported difference in body mass index (BMI) was due to the method of data collection (actual heights and weights vs self-reported heights and weights). We decided to conduct our own telephone survey in February 1997 to compare self-reported heights and weights with the actual heights and weights measured in the clinics.
In 1993 Michigan ranked sixth in the United States in the percentage of overweight persons in the population with a prevalence of 29%, according to the Michigan Behavioral Risk Factor Status Survey (MBRFS),1 which is a telephone survey conducted each year by the state health departments and the Centers for Disease Control and Prevention. The results of the MBRFS random digit dialing survey of 2000 adults differed from a height and weight study conducted in UPRNet practices during the spring and summer of 1996. The actual office-based height and weight study showed that 53% of the population was overweight. Because of the disparity in the results of these 2 studies in the same state, we wondered if self-reported heights and weights are grossly inaccurate or if the population in primary care offices is different from the general population of Michigan. Several studies3-8 of telephone-reported weights confirm that as the survey population becomes heavier, self-report is less accurate.
The original heights and weights study took place in 19 UPRNet practices and 2 urban-based ambulatory family practice clinics located in the Lower Peninsula. The telephone follow-up study was conducted in 2 UPRNet practices. A total of 150 patients 18 years and older were randomly selected from the daily clinic patient schedule during a 1-week period. At the time of their office visits, patients were asked to give their consent to participate in a study of health behaviors. Patients were weighed and measured during the visit. Within a week of their office visits, patients were phoned and asked to report their current height and weight. Several health behavior questions from the MBRFS were also posed. Three attempts were made to contact each patient. The University Committee on Research Involving Human Subjects of Michigan State University approved the study protocol.
The results were analyzed using a paired t test to determine whether the difference between actual and reported heights and weights was statistically significant. Mean BMI was calculated using Statview.4,5
Of the 150 patients who consented to participate during their office visits, 89 (59%) responded to the questions asked in the phone interview. Sixty-one of the respondents were women. Twenty-two patients who consented to participate refused to answer the phone survey questions. Thirty-eight of those who gave their consent could not be reached by telephone after 3 attempts. One patient was on a weight-reduction program at the time of the study; we eliminated her data.
Of the 89 patients weighed in the office, 57% of the men and 48% of the women were overweight. When contacted by telephone, 50% of the men and 46% of the women reported weights in excess of normal for their age and sex. Women underreported their weight by 6 pounds (P <.001) and reported their height accurately. Men underreported their weight by 6.2 pounds (P <.001) and overreported their heights by one-half inch (P = .0088). The average BMI reported by the women was 28.75, actual BMI of 30.09 (difference of 1.34); reported BMI for the men was 28.66, actual BMI of 29.98 (difference of 1.32). The difference between self-reported BMI and actual BMI for the study group as a whole was statistically significant (P=.014). There was no significant statistical difference between how the men reported BMI and how the women reported it (difference of .02).
Compared with the MBRFS study population, the UPRNet sample included more people of lower socioeconomic status and was almost totally non-Hispanic white. People enrolled in the MBRFS were more representative of the population of the state economically and racially.
Other studies of self-reported weight indicate that overweight men and women generally underreport their actual weights. A surprising result of our study was the number of people who reported a weight that indicated they were underweight, rather than (their actual) normal weight. This difference in self-report and actual healthy weights was unexpected, since it has been reported that the least accurate self-reports come from heavier individuals.3
In other telephone studies, it was found that heavier women were the least accurate in self-report.3 Our study found that men and women underreported their weights approximately equally and that men also overreported their heights.