- Correspondence to: Dr N S Crowcroft Epidemiology Department, Scientific Institute of Public Health–Louis Pasteur, 14 Juliette Wytsmanstreet, Brussels B-1050, Belgium.
- Accepted 11 March 1997
Objective: To describe risk factors for infantile colic.
Design: Questionnaire administered by health visitors.
Subjects: Mothers of 76 747 infants born between 1 August 1975 and 31 May 1988, interviewed when the infant was 1 month old.
Main outcome measures: Reporting of infantile colic and its duration; weight of infant, feeding, state of the home, socioeconomic characteristics of the parents, parents' age, and mother's parity.
Results: The odds of reporting infantile colic were increased with breast feeding (odds ratio of breast v bottle feeding 1.35 (95% confidence interval 1.28 to 1.43), increasing parental age, lower parity, increasing parental age at leaving full time education, and more affluent homes and districts of residence. In a logistic regression analysis, mother's age and parity and socioeconomic factors remained the most important risk factors for the reporting of infantile colic (each P < 0.005), and the effect of breast feeding was attenuated (odds ratio of breast v bottle feeding 1.09 (1.02 to 1.15).
Conclusion: At a population level, dietary factors contribute little to mothers' reporting of infantile colic, and dietary change should not be the primary intervention.
Infantile colic is poorly defined but commonly reported and causes parents appreciable distress
Often the first advice that parents receive is to change the infant's diet
Slightly higher rates of reporting of infantile colic are found when infants are breast fed than bottle fed, so formulas based on cows' and hence allergy to cows' milk protein are unlikely to be important causes of infantile colic
Social factors are most important, with older primigravid mothers who have a non-manual occupation and who stayed in full time education longest reporting the highest rates of colic in their infants
Dietary change should not be the first intervention for colicky babies
Infantile colic is a syndrome characterised by paroxysmal, excessive, and inconsolable crying without identifiable cause in a healthy infant. It is also called persistent crying in infancy and three month colic because it usually disappears by three months of age. Infantile colic is common.1 Estimates of cumulative incidence have varied, depending on the case definition and the period of follow up, from 10% to 40%.2 3 It is usually self limiting, without long term adverse consequences, but caring for an infant with colic can be distressing and frustrating for parents. At the extreme its effects on the parent-infant relationship may be sufficient to disrupt the infant's development,4 5 6 and it may increase the risk of child abuse.7 Research has not established the aetiology or best management of infantile colic, and infant crying was identified as a priority area in the NHS Research and Development Programme.8
The Sheffield child development study was designed to identify infants at high risk of sudden infant death but included questions on parental reporting of colic.9 10 We aimed to identify risk factors for parental reporting of infantile colic during the first month of life. Clarifying the aetiology may indicate possible approaches for prevention and for clinical management.
Subjects and methods
In the Sheffield child development study, all parents of infants born in Sheffield between 1975 and 1995 were interviewed by a health visitor at the routine visit when the infant was about one month of age. Our study used information collected from 1975 to 1988 because data were computerised only for this period.
A detailed questionnaire was completed on a range of subjects which included characteristics of the family, feeding practices, and the behaviour of the infant. Ascertainment of colic relied on parental reporting, as in other studies.3 11 Health visitors asked mothers whether or not the baby was currently colicky as well as the number of days the baby had been colicky since birth. The questions on colic formed part of a short series of questions within the questionnaire which were all in the same format and which concerned various minor symptoms. Colic was not formally defined for the study, but the health visitor could offer clarification if the mother asked for it, as at any routine visit.
The age at interview ranged from two to more than 99 days. The analysis was restricted to babies aged between 24 and 37 days, representing 87.5% (67 172/76 747) of the births and 92% of those for which data on colic were available (67 172/72 995).
Our analysis explored hypotheses identified from the literature1 2 7 11 relating to diet and social and economic factors and tested the null hypothesis that babies who are reported to be colicky are no different from babies who are not.
Univariate analyses were performed using SAS version 6.09.12 Relations between categorical variables were examined using the χ2 test or, for ordered categorical variables, the χ2 test for trend. Stratified tests for trend were carried out in EpiInfo version 5. Continuous variables were compared by using the large sample normal test for the differences between means and Pearson's correlation coefficient.13 The logistic regression was carried out in GLIM4.14 Variables were included if there was a significant association with colic in the univariate analysis (at a significance level of 0.05) or if they were likely confounders. The dependent variable was a dichotomous measure of parental reporting of colic (current and past) versus no history of colic. Explanatory variables were tested in a forward stepwise regression analysis using a χ2 test of heterogeneity or for trend where appropriate.
In total, 12 277 infants were reported to have been colicky at some stage in the first month, a cumulative incidence in the first month of life of 18.3% (12 277/67 127), and 8251 infants were recorded as currently being colicky, a point prevalence of colic at one month of 12.3% (8251/67 127). Colic and sex of the infant were not related, with a prevalence of 18.4% in boys and 18.2% in girls. The prevalence of colic in babies described as “demanding” was 38.9% (3105/7983) compared with 15.5% (8942/57692) in those described as contented or never crying.
Colicky babies weighed more at birth, gained more weight, and weighed more at 1 month of age (table 1). After adjusting for maternal education, parity and maternal age the differences in birth weight (21.33 g, 95% confidence interval 11.47 g to 21.21 g) and weight gain as a percentage of birth weight (0.44%, 0.21% to 0.67%) were trivial although they remained significant (P < 0.0001).
Information was recorded about the longest period that the baby was reported to spend continuously asleep or awake. Colicky babies had significantly shorter periods of “longest continuous time asleep” (mean difference 14.8 (95% confidence interval 13.0 to 16.7) minutes, P < 0.0001) and longer periods of “longest continuous time awake” (29.1 (27.1 to 31.0) minutes, P < 0.0001).
Babies who had been or were being breast fed were significantly more likely to be reported as colicky (χ2=198.4, df=6, P < 0.0001) (table 2). The prevalence of past or current colic in exclusively breast fed babies was 20% compared with 16% in exclusively bottle fed babies, a relative risk of 1.62 (1.55 to 1.71). There was also a significant trend for breast fed babies to have a longer reported duration of colic than bottle fed babies (χ2 test for trend comparing breast only and bottle only, P < 0.0001) (table 2).
Of the infants who were bottle fed, colicky babies were more likely to have had a change in type of formula feed and to have had more changes in type of formula than non-colicky babies. Of the infants who had one or more changes in formula, colicky infants had a mean of 1.2 changes, compared with 1.1 for non-colicky babies (P < 0.0001).
Table 3 shows that there was a non-linear relation between maternal age and past or present colic, the unadjusted prevalence of colic being highest among the offspring of mothers aged 30-34 years. The prevalence and odds of reported colic fell progressively with increasing parity.
Health visitors subjectively assessed some characteristics of the environment of the child on a five point scale, including the type of neighbourhood and the state of repair of the house, furnishings, and equipment. The prevalence of reported colic showed a trend of increasing colic with more affluent neighbourhood and better state of repair of the home (P < 0.0001).
Educational achievement was examined as a proxy measure for current social class, as parental occupation was not recorded in the Sheffield child development study until 1983. Increased rates of reporting were observed in mothers who were older when they left full time education (table 3). From 1983 to 1988, the prevalence of reported colic was higher where mothers or fathers had “white collar” occupations (χ2 = 166.3, df = 8, for mother's or father's occupation, P < 0.0001) (table 4).
The logistic regression analysis was performed to control for confounding and to estimate the strength of the risk associated with significant explanatory variables. Two logistic regression models were created, one using all the data from 1975 to 1988 (table 3) and the other including only data from 1983 to 1988, when parental employment was recorded (not shown). Excluding the missing data reduced the number of infants to 56 949/67 172 (85%) for the period 1975-88 and to 25 952/33 554 (77%) for the period 1983-8.
The odds ratios for current feeding method shown in table 3 use exclusive breast feeding as the reference category. Bottle feeding was associated with reduced odds of reporting colic compared with breast feeding, after controlling for potential confounders (odds ratio 0.92). In the 1983-8 data, the ratio of the odds of reporting colic by mothers who bottle fed to that of those who breast fed was 0.93 (0.85 to 1.02) maternal age, parity, education and occupation were controlled for. Changing from breast to bottle significantly increased the odds of reported colic in comparison to exclusive breast feeding in both models (odds ratio 1.13 (1.06 to 1.20) v 1.13 (1.02 to 1.25).
After adjustment for mother's and father's age and educational achievement, mother's parity, and feeding method, a strong trend remained of greater odds of reported colic in houses judged to be in a better state of repair, but the independent effect was weak. Maternal education was a stronger independent correlate of reported colic than father's school leaving age (table 3). The relation between mother's age and infantile colic was strengthened after adjustment, with an increase in the odds of colic in older mothers (table 3). The trend of lower odds of colic with increasing parity was little affected by adjustment.
This large, population based survey confirms that infantile colic is commonly reported and associated with an increased chance of the mother finding the infant demanding and changing the pattern of feeding. However, feeding method did not emerge as an important determinant of the incidence or duration of colic after allowance for the stronger influences of maternal age, parity, and socioeconomic circumstances.
As these data cover a virtually complete set of births between 1975 and 1988, the problems of selection bias that are seen in many other studies of infantile colic have been avoided.15 Furthermore, the size of this study gives it the power to exclude with confidence all but the most subtle associations. Even small differences between groups that may not be clinically important and that may be the result of residual confounding are highly statistically significant.
Cases of colic that developed after the infant was older than 1 month were not ascertained, but in 90% of cases, colic starts in the first month of life.16 In this study, most babies were seen close to 1 month of age, and babies who were seen before 24 days were excluded from the analysis, so most colicky babies should have been identified.15 17 However, babies who present before 1 month of age may differ systematically from those who present at a later stage, in having more severe colic.16
The study deals only with what is reported and so no comment can be made about whether the infants labelled as colicky actually cried more than other infants. This does not represent a problem of interpretation for the application of the findings to clinical practice since health professionals usually have to advise on the basis of how infants are reported to behave. This may, however, have introduced imprecision into the ascertainment of colic at 1 month, which would have the effect of underestimating differences between colicky and non-colicky infants.
Interpretation of findings
A unifying interpretation of the main findings of this study is that the reporting of infantile colic by mothers is largely a social phenomenon. Mother's age and parity were the most important factors influencing the reporting of infantile colic, followed by the additional effect of socioeconomic factors. These findings are consistent with the hypothesis that parent-child interaction may be important in determining the reporting of colic. This interaction is influenced by a range of other factors including parents' expectations and interpretations of their infant's behaviour and the availability and use of coping mechanisms. The infant's sex did not influence the rate of reporting of colic, and this may argue against there being a profound psychosocial bias acting on the mother, since parents respond differently to male infants.18 The design of this study, with quantitative information collected by questionnaire, means that it is not possible to examine the underlying causes of the reporting behaviour in any depth.
Babies who were reported to be colicky were also reported to sleep less and have much longer periods of wakefulness than non-colicky babies. As the amounts of sleep and wakefulness were not objectively measured, this finding could be subject to bias since mothers of colicky infants may be more likely to perceive that their infants do not sleep, or mothers of infants that do not sleep to perceive that their infant is colicky. However, babies who are reported to be colicky may indeed sleep less and may behave differently in other ways from non-colicky babies.
Mothers of breast fed infants reported colic at a higher rate than did mothers of bottle fed infants, but this effect was greatly reduced after adjustment for potential confounders in the logistic regression analysis. The confounder which had the greatest effect in reducing the independent effect of breast feeding was social class (measured by the proxies maternal education or occupation). As such social factors are poorly measured, their effects may be underestimated,19 so the independent effect of breast feeding after adjustment may be the result of underadjustment for socioeconomic factors. If the type of milk feed influences the occurrence of infantile colic, its effect is either very small or takes place in a small minority of infants. Formula milk is most frequently derived from cows' milk. Since bottle feeding was associated with a lower odds of reporting colic than breast feeding, allergy to cows' milk protein is unlikely to account for the majority of reported colic.
The differences between colicky and non-colicky infants in birth weight and weight gain were statistically significant but were reduced by adjusting for confounders and in any case would not be regarded as clinically important. They were in the direction of colicky infants thriving.
At a population level, dietary factors contribute little to mother's reporting of infantile colic, and dietary change should not be the primary intervention. Future qualitative research is indicated to explore the social factors which influence the reporting of colic and interventional research should focus on both practical and psychosocial aspects in examining ways to alleviate the impact of persistent infant crying on families.
We thank Dr Elizabeth Taylor (University Department of Paediatrics, Sheffield) and Dr Robert Carpenter (London School of Hygiene and Tropical Medicine) for allowing access to the records of the Sheffield child development, and the health visitors of Sheffield, who collected the data.