Wald, Ellen R.*; Jagodzinski, Tanya D.†; Moyer, Stacey C.L.*; Wald, Arnold*; Eickhoff, Jens C.*; Edmonson, M. Bruce*
Constipation is a common complaint in children, accounting for an estimated 1% of all pediatric office visits (1,2) and 20% of visits to a pediatric gastroenterologist (3,4). Unfortunately, constipation can become chronic in many children before the problem has been recognized by a parent or a primary care physician. It has been estimated that at least 50% of parents of children with constipation do not recognize that their children are constipated (5). Because retentive fecal soiling may be a consequence of constipation in early school-age children, it is greatly desirable to avoid constipation; appropriate guidance from a primary care provider may prevent the development of chronic constipation (6). Accordingly, the development of a questionnaire that may alert parents and primary care providers to the presence of subclinical or unrecognized constipation would be useful. The provider could easily use such an instrument to screen patients in susceptible age groups and be prompted to explore their bowel habits more thoroughly if the screening questionnaire were positive.
We previously characterized the bowel habits of >1142 primarily middle class white children 5 to 8 years of age presenting to their primary care practitioner for a health maintenance visit (7). The validity of information concerning bowel habits of children obtained by parental recall is not well established. The purpose of the present study was to correlate recall data from a bowel habit questionnaire (BHQ) with prospectively obtained data on bowel habits from a 14-day calendar diary. In addition, our objective was to determine whether the BHQ would predict the development of medically significant constipation (MSC) 1 year later. The elements that comprised the BHQ were adapted from previously published consensus criteria for constipation in children (8,9).
MATERIALS AND METHODS
The receptionist or nursing staff at each of 11 general pediatric clinics in Dane County, Wisconsin, identified children ages 5 to 8 years as they arrived for a health maintenance visit. The total population of 5 to 8 year olds within the 11 clinics was 8810. A packet containing the consent form and BHQ was distributed to the parents of all of the eligible children during the check-in process. If family members chose to participate, then they were instructed to complete the attached consent/authorization form and the BHQ and to return both to a designated receptacle in the waiting room. The packet also contained a stamped envelope that families could return by mail if they chose to complete the BHQ at home.
BHQ: The Questionnaire
The questionnaire included demographic data, questions about the age at which toilet training was completed, and the child's current bowel habits. In addition, there were questions relating to MSC, which was defined as previous medical encounters about constipation or frequent stool accidents, or use of enemas or suppositories more than once per year or regular use of laxatives or stool softeners. This definition reflected the fact that the parent had either sought care or instituted treatment for perceived constipation. The BHQ classified children as constipated if their total score was at least 2, after assigning 1 point for each of the following: bowel movement <3 days/week, stool accidents at least once in the previous 2 weeks, straining at stool >25% of the time, avoidance >25% of the time, discomfort with defecation >25% of the time, or passing “very large stools” >25% of the time. This definition (in which palpation of a large fecal mass was replaced by “straining” on defecation) was adapted from previous consensus statements (8,9).
The consent forms and BHQs were retrieved each week by the study coordinator. The following week, families of selected subjects received a bowel habit diary by mail. Families were asked to complete a daily record of their child's bowel habits for 14 days (see the BHQ and diary in the online-only appendix at http://links.lww.com/MPG/A53). A stamped return envelope was provided. If the diary was not returned within 3 weeks, then a telephone call was made to remind families. A second diary was sent in 1 month if there was no response. One year later, families who returned the diary and did not report MSC on the first BHQ received a second BHQ in the mail along with a telephone reminder to complete the second BHQ. Families received $25 for completing the diary and $25 for completing the second BHQ. The study was approved by the institutional review board at the University of Wisconsin School of Medicine and Public Health.
Sampling From Clinics
From our previous study, we anticipated the following distribution of scores for BHQs: 70% with a score of 0, 20% with a score of 1, and 10% with a score of at least 2 (7). To maximize sampling efficiency, we planned to solicit diaries from 200 children who had a BHQ score of 0, 200 with a score of 1, and 100 with score of more than or equal to 2. Furthermore, to achieve nearly equal enrollment from each practice, diaries were not sent to additional children with a BHQ score of 0 once there were 30 children with that score in a given practice. Families were enrolled between July 17, 2006 and May 13, 2008. The last diary was completed and returned on June 18, 2008. The last follow-up BHQ was received on June 12, 2009.
Bowel Habit Diary
Bowel habits as recorded on the BHQ were compared with the diary. The choices for frequency of defecation on the BHQ were daily, every other day, <3 times per week, or unknown. The following definitions were used for the diary:
1. A child was considered to have had a daily bowel movement if there was at least 1 bowel movement per day on 12 or more days during the 14-day period.
2. A child was considered to have had a bowel movement every other day if there was at least 1 bowel movement per day on 6 to 11 days during the 14-day period.
3. A child was considered to have had <3 bowel movements per week if there were <5 bowel movements during the 14-day period.
The proportion of stools described as “very large/huge” was calculated directly from the 14-day period. One point was assigned if >25% of stools were considered “very large.” The proportion of days with straining, discomfort, and avoidance was calculated directly from the 14-day period. A point was assigned for any characteristic that was present >25% of the time. The number of stool accidents each day was recorded. If there was at least 1 accident in 14 days, then 1 point was assigned.
Descriptive statistics were generated to summarize the data. Categorical data were summarized as proportions and percentages. Continuous data were summarized and reported as means and their standard deviations. A 2-sample t test was used to compare continuous variables between groups. Receiver operating characteristics (ROC) analysis was conducted to determine the threshold value of the BHQ total score for defining constipation. The sensitivity levels for the BHQ total score threshold values 1, 2, 3, and 4 were 74%, 60%, 41%, and 28%, respectively, whereas the corresponding specificity levels were 65%, 83%, 94%, and 96%, respectively. There was a substantial decrease in sensitivity when changing the threshold from 2 to 3 with only a small increase in specificity. Hence, based on the results of the ROC analysis, the optimal threshold for determining disease status was determined to be 2, that is, the BHQ classified children as constipated if their total score was at least 2. These were the same criteria used by other investigators (8,9). The sensitivity, specificity, and positive and negative predictive values for each element of the BHQ were computed and reported along with their corresponding 95% confidence intervals (CIs). The normal approximation method was used to compute the 95% CIs. The bootstrap method was used to construct the 95% CIs of the positive and negative likelihood ratios.
Figure 1 shows a flow diagram for patient progress through the study. The BHQ was distributed to 1193 families; 416 returned the original BHQ. Two hundred sixty-five (63.7%) children had a score on the first BHQ of 0, 60 (14.4%) had a score of 1, and 91 (21.9%) had a score of at least 2. Fifty-seven of these 416 (13.7%) children had previous or current MSC; 9 (15.8%) had a score of 0 on the BHQ, 5 (8.8%) had a score of 1, and 43 (75.4%) had a score of at least 2. Diaries were sent to 360 families (including those with MSC) and were returned for 269 children. Data on these children became the basis for paired comparisons to validate the BHQ against the diary.
Demographic information for the 269 children in the validation sample is shown in Table 1. These children were balanced for sex. The racial distribution of subjects was similar to that of Dane County. Seventy-nine percent of the parents held baccalaureate or higher degrees, reflecting the high educational level in the area surrounding the University of Wisconsin. The ages at completion of toilet training were different for boys and girls (37.3 and 33.2 months, respectively), and were higher than expected. The mean age of toilet training for 42 of 57 children (who returned the diary) with MSC was 39.6 months (± 11.3 standard deviation) in contrast to a mean age of 34.6 months (± 8.2) for 220 (of 262) children without constipation (P < 0.001).
The symptoms of constipation recorded in the diary are shown in Figure 2. Straining was the most common symptom followed by large stools; in contrast, infrequent bowel movements were recorded for only 6% of children. The sensitivity and specificity of the individual components of the BHQ as compared with the diary are shown in Figure 3. The 2 most sensitive components were stool infrequency and accidents. The overall sensitivity of the BHQ for constipation was 59.6% (95% CI 46.7%–71.4%) and specificity was 82.6% (95% CI 77.0%–87.1%). A total of 54 of 269 (20.1%) children met criteria for constipation by the data in the 14-day diary report, including 32 of 45 (71.1%) who had MSC on their first BHQ and returned their diary.
One year later, based on caregiver responses for 210 (93.8%) of 224 subjects, 11 children (5.2%) had developed MSC for the first time. Seven (63.6%) of these children had BHQ scores of at least 2 on their first BHQ (sensitivity 63.6%, 95% CI 35.4%–84.8%). Among children who did not develop MSC, most (170/199) had BHQ scores <2 (specificity 85.4%, 95% CI 79.9%–89.7%). The positive predictive value for the first BHQ for subsequent MSC was 19.4 (95% CI 9.8–35.0), the predictive negative value was 97.7 (95% CI 94.2–99.1), and the likelihood ratios positive and negative were 4.4 (95% CI 2.5–7.6) and 0.42 (95% CI 0.19–0.93), respectively.
The ability of the BHQ to identify constipation in children was found to be only moderately sensitive but somewhat more specific when compared with the criterion standard 14-day diary. Children whose original BHQ score was at least 2 were approximately 4 times more likely to develop MSC during the following year than children with a score <2. Most children (81.6%) with a BHQ score of at least 2 did not develop MSC. Nonetheless, we gained considerable insight about constipation in children in the course of the study. In our sample, 13.7% of children had received medical attention or treatment for constipation by early school age. An even larger percentage of children (20.1%) had subclinical constipation as indicated by the diary. For most of these children, constipation remained subclinical as demonstrated by a 1-year follow-up, which showed that medical advice or treatment had been sought for only 11.4% of children with parental responses on BHQ indicating constipation.
The most sensitive item on the BHQ was infrequency of bowel movements, which had a sensitivity of 100% and a specificity of 92%. Infrequent bowel movements were reported in only 6% of the subjects in the present study. Accidents, which also had a relatively high sensitivity and specificity, were reported in only 8% of children. Straining, which was reported in 26% of children, had a sensitivity of only 49% and a specificity of 86%.
Our previous investigation, conducted in Pittsburgh, Pennsylvania, evaluated bowel habits in a large (1142 children in the same age group) and diverse group of healthy children (7). Compared with the present study, it showed lower baseline rates for several characteristics of defecation, suggesting that the present study population was probably enriched for children with constipation (7). Only 10% of children in the Pittsburgh study had a BHQ of at least 2 compared with 22% of 459 children in the Wisconsin study (P < 0.001). One explanation for the discrepancy between studies likely relates to the method of enrollment. Because no follow-up was planned after completion of the BHQ in the Pittsburgh study, no personally identifying information was collected and informed consent was not required (7). In the present study, validation of the BHQ required that we be able to follow the group of children for whom the first BHQ was recorded. The necessity of obtaining informed consent likely selected for families most interested in participating in the study and may have altered the composition of the study group and, perhaps to some extent, its generalizability. It is important to validate the Rome III and Paris Consensus on Childhood Constipation criteria in unselected populations of children in both general practice and research settings (10). This is a challenge in the environment of stringent regulation of medical research, even for minimal-risk studies such as ours.
Similar to other functional bowel disorders for which there is not a criterion standard, the definition of constipation and criteria for its identification have been controversial. Most efforts to validate definitions have studied populations of children with a chief complaint of constipation who were evaluated in digestive disease clinics or referred by their primary care providers with constipation as the chief complaint (7). In 2 studies performed in Italy, primary care providers completed a detailed questionnaire that recorded symptoms, signs, and laboratory tests needed to satisfy the Rome II criteria (1,2). In contrast, our approach was to devise a score, by adapting criteria developed by the Paris consensus group and endorsed by Rome III, that would prospectively identify children with subclinical constipation (ie, not recognized by families) (10,11). Such an instrument would be useful to the primary care practitioner to identify (and perhaps preemptively treat) these children in the hopes of avoiding the development of “encopresis” or retentive soiling.
The Rome diagnostic criteria for functional constipation are nearly identical to the Paris Consensus on Childhood Constipation Terminology (9,10). Both require the inclusion of at least 2 of the following features in a child with a developmental age of at least 4 years: 2 or fewer defecations in the toilet per week, at least 1 episode of fecal incontinence per week, a history of retentive posturing or excessive volitional stool retention, history of painful or hard bowel movements, presence of a large fecal mass in the rectum, and presence of large-diameter stools that may obstruct the toilet. Because a rectal examination could not be part of a screening test, our score omitted this item.
A potential limitation of the present study is that the caregiver who filled out the BHQ was not specifically instructed to solicit his or her child's input when completing either the BHQ or the 14-day diary. This may be an important omission if parents are not always aware of the toileting practices of their children in the age group 5 to 8 years. We speculate that this omission may account, at least in part, for the limited sensitivity of the BHQ. Another legitimate question is the accuracy of recall, that is, even if parents were generally knowledgeable about the bowel habits of their children, how accurate would their recall be? Recall may be inadequate to reflect actual information concerning bowel habits. Alternatively, recalled data, especially when supplemented by input from the child, may be accurate for some characteristics of defecation (11). Moreover, there may be substantial fluctuation in bowel habits from week to week that precludes the comparison of a diary with a recall questionnaire. This is less likely, because 2 weeks is a reasonable period to reflect actual variation that may exist.
Although imperfect, we believe that the BHQ has substantial potential as a clinically useful instrument. Our intent in future studies is to ensure that parents consult their children while completing the questionnaire. It is likely that the performance characteristics of the BHQ will improve under these circumstances. Identifying children with subclinical constipation is an important part of health maintenance.
1. Miele E, Simeone D, Marino A, et al. Functional gastrointestinal disorders in children: an Italian prospective survey. Pediatrics
2. Primavara G, Amoroso B, Barresi A, et al. Clinical utility of Rome criteria managing functional gastrointestinal disorders in pediatric primary care. Pediatrics
3. Caplan A, Walker LS, Rasquin A. Development and preliminary validation of the questionnaire on pediatric gastrointestinal symptoms to assess functional gastrointestinal disorder in children and adolescents. J Pediatr Gastroenterol Nutr
4. Partin JC, Hamill SK, Fischel JE, et al. Painful defecation and fecal soiling in children. Pediatrics
5. Loening-Baucke V. Encopresis and soiling. Pediatr Clin N Am
6. Borowitz SM, Cox DJ, Kovatchev B, et al. Treatment of childhood constipation by primary care physicians: efficacy and predictors of outcome. Pediatrics
7. Wald ER, DiLorenzo C, Cipriani L, et al. Bowel habits and toilet training in a diverse population of children. J Pediatr Gastroenterol Nutr
8. Rasquin A, Di Lorenzo C, Forbes D, et al. Childhood functional gastrointestinal disorders: child/adolescent. Gastroenterology
9. Benninga M, Candy DCA, Catto-Smith AG, et al. The Paris Consensus on Childhood Constipation Terminology (PACCT) group. J Pediatric Gasteroenterol Nutr
10. Boccia G, Manguso F, Coccorullo P, et al. Functional defecation disorders in children: PACCT criteria versus Rome II criteria. J Pediatr
11. van der Plas RN, Benninga MA, Redekop WK, et al. How accurate is the recall of bowel habits in children with defaecation disorders? Eur J Pediatr