Objective. Although early otitis media (OM) onset predicts later recurrent and chronic OM, little research has been directed at illuminating the role of prenatal exposures in early OM. This prospective study examined prenatal, innate, and early environmental exposures associated with acute otitis media (AOM) onset and recurrent OM (ROM) by age 6 months.
Design and Methods. Prospective study of 596 infants from a health maintenance organization followed from birth to 6 months. Mothers completed monthly forms on prenatal exposures (diet, medications, and illnesses) and infant risk factors (eg, smoke exposure and child care) during pregnancy and until infants were 6 months old. Urine samples were collected when infants were 2 months of age and analyzed for cotinine and creatinine. Physicians and nurse practitioners examined infants at each clinic visit and completed standard ear examination forms.
Results. Thirty-nine percent had an episode of AOM and 20% had ROM by age 6 months. Using Cox's regression models to control for confounding, respiratory tract infection (relative risk [RR] 7.5), day care (RR 1.7), >1 sibling (RR 1.4), maternal, paternal, and sibling OM history (RR 1.6, 1.5, and 1.7, respectively) were significantly related to early OM onset. ROM was related to respiratory tract infection (RR 9.5), day care (RR 1.9), conjunctivitis (RR 2.0), maternal OM history (RR 1.9), and birth in the fall (RR 2.6). Among prenatal exposures, only high prenatal dietary vitamin C intake was significantly inversely related to early AOM with univariate but not multivariate analysis.
Conclusion. Prenatal factors were not linked to early AOM onset with multivariate analysis, but environmental and innate factors play an important role in early AOM onset. Strategies to reduce exposure to environmental variables could reduce rates of early AOM, which could potentially result in declining rates of ROM and chronic OME.
- OM =
- otitis media •
- OME; otitis media with effusion; AOM =
- acute otitis media •
- HMO =
- health maintenance organization •
- CCR =
- cotinine:creatinine ratio •
- ROM =
- recurrent otitis media •
- ARR =
- adjusted relative risk •
- RR =
- relative risk •
- RTI =
- respiratory tract infection •
- CI =
- confidence interval
Children who experience otitis media (OM) onset in the first few months of life are at greater risk of chronic otitis media with effusion (OME) and recurrent OM than children who have later onset.1–4 Although an immature immune system and anatomic factors may contribute to the overall high OM incidence in infancy, prenatal and early environmental exposures may also play a role. Research has focused on the prenatal and neonatal periods to identify factors that increase the risk of childhood cancer5–8 and determinants of adult diabetes, hypertension, and cardiovascular disease.9–11 However, little research has been devoted to prenatal influences on OM onset, although OM incidence is highest in the first 2 years of life and it is the most commonly diagnosed childhood disease.12 This prospective study was designed to identify prenatal, infant, and early environmental exposures that predict acute OM (AOM) onset and recurrent OM in the first 6 months of life. We hypothesized that infants would be at greater risk of early OM if their mothers had low prenatal intakes of zinc, iron, vitamins important in immune function, or high alcohol intakes. Infant characteristics and early factors hypothesized to confer greater risk of early OM included low birth weight, early gestational age, day care attendance, and urine cotinine levels indicative of exposure to passive smoke. Several of these areas of investigation have been identified as research priorities for OM.13 ,14
Pregnant women who received health care at Group Health Inc (now HealthPartners), a large health maintenance organization (HMO) in the twin cities metropolitan area were enrolled in the Early Otitis Media Study between October 1991 and August 1994. A subset of these women (42%) were also participants in the Diana Project,15 a study of the influence of preconception and prenatal factors on infant birth weight. Women were eligible if they were at least 18 years old and received prenatal care at a staff model clinic. Infants were eligible if they were born during the study enrollment period and received care at an HMO staff model clinic; they were excluded if they had a craniofacial anomaly predisposing them to OM (eg, cleft palate, or Down syndrome). Between 1991 and 1993, Diana participants were contacted by phone during the sixth month of pregnancy and invited to enroll in the study; other HMO enrollees were recruited by letter during 1993 and 1994 and urged to contact study staff if interested in a study of child health, the same method used to recruit women for the Diana Study. All women provided written consent for themselves and their infants; Diana participants also allowed access to their data from that study. The study has been reviewed and approved annually by the institutional review boards of the University of Minnesota and HealthPartners.
All women completed monthly questionnaires from the last trimester of pregnancy until the infant was 6 months old; Diana participants also completed monthly questionnaires during the first two trimesters. Prenatal questionnaires pertained to use of tobacco and alcohol, maternal symptoms and illnesses, and use of dietary supplements, medications, and recreational drug use. Dietary intake was estimated with the Willett Food Frequency questionnaire,16 and vitamin and mineral supplements were coded by brand, compared with a master list of constituent quantities for each brand, and converted to average daily intake for each constituent. Questionnaires about the infant's first 6 months of life included symptoms, illnesses, known and possible risk factors for OM (eg, day care and related variables, family history of recurrent or chronic OM, method of feeding, introduction of foods or supplements, siblings, pets, parental smoking and other smoke exposure variables, and 3-day records of infant dietary intake at 1, 3, and 5 months of age). Dietary data were reduced to intake of individual constituents, and divided into quartiles for statistical analysis. Trained study staff abstracted neonatal and infant health variables from the clinic record.
Cord bloods (n = 425) were collected at delivery, sera were assayed for antibody levels to seven Streptococcus pneumoniae serotypes, and relationships between cord blood antibody levels and age of OM onset were explored. These analyses included only a subset of the cohort, and have been reported elsewhere.17 Therefore, antibody levels were not included as a variable in these analyses. Urine samples were collected at the 2-month well-child examination, stored at −20°C, and assayed at the American Health Foundation (Valhalla, NY) for cotinine and creatinine. Because urine samples were random rather than 24-hour collections, cotinine levels (ng/mL) were standardized to creatinine levels to adjust for urine concentration and reported as nanograms per milligram (ng/mg) as cotinine:creatinine ratios (CCR). Cotinines were assayed in duplicate by radioimmunoassay with specific antisera produced in rabbits, using a modification of the method of Langone et al,18 and the average of the duplicate values was used. Fifty-two samples were divided and assayed as blind duplicates. Intraassay and interassay variation was 7%, with a sensitivity of 2 ng/mL. Creatinines were measured by dry chemistry methods on a Kodak Ektachem 500 analyzer (Johnson and Johnson Clinical Diagnostics, Rochester, NY).
All infants were enrolled at birth and followed in the HMO for 6 months. At each clinic visit the physician or pediatric nurse practitioner examined the child with pneumatic otoscopy, using a standard protocol, and recorded tympanic membrane findings, symptoms, and diagnoses on a standard one-page data form. Tympanometry was also performed at 2, 4, and 6 months, and when needed for an equivocal middle ear diagnosis. If an ear form was not completed for a clinic or urgent care visit, study staff abstracted data from the medical record for that visit. Parents signed a medical release for infants who left the HMO and their OM histories were abstracted from the clinic record of the new provider.
Study outcomes included early AOM, defined as a physician-diagnosed episode of AOM during follow-up from birth to 6 months, and recurrent OM (ROM), defined as 2 or more episodes of OM (AOM or OME) by 6 months of age. A new episode was defined as AOM or OME in either ear after a normal middle ear examination, or an episode of AOM ≥21 days after a diagnosis of AOM or OME. OM treatment was determined by the physician, but the HMO standard of care was to use first-line antibiotics (eg, amoxicillin, or trimethoprim sulfamethoxazole) for new OM episodes. The protocol did not require tympanocentesis to determine the presence of fluid or to identify bacterial pathogens when OM was diagnosed. To assess reliability of the middle ear diagnosis, validated otoscopists performed interobserver testing with a sample of 20 clinicians. All clinicians' diagnoses were compared with a computer algorithm in which OM was considered present if two or more of the following tympanic membrane findings were abnormal: color (not gray); position (full, bulging, or retracted); appearance (opaque or dull); or mobility (decreased or absent).
Incidence rates (proportion of children experiencing AOM) were based on the number of infants followed from birth to 6 months (denominator) and the number followed for that period who experienced either AOM or ROM (numerator). Episodes per child-year of follow-up included all infants with any follow-up in both the numerator and denominator. Two CCR cut points were examined as indicators of infant exposure to passive smoke, ≥40 ng/mg and >100 ng/mg.19 ,20 Relationships between risk factors and AOM and ROM by age 6 months, differences between participants and withdrawals, and Diana and other HMO participants were explored with univariate analysis by using χ2 and t tests of independent means.21 Adjusted relative risks (ARRs) for AOM and ROM were derived from Cox's regression models including all factors significant (P < .05) with univariate analysis.22 Cox's regression allows for estimation of relative risks (RRs) independent of length of follow-up if there are sufficient events to estimate these parameters. For AOM, these RRs were compared with those resulting from a forward stepping (P < .05, to enter), parsimonious model. In these regression models, time variable factors such as day care, respiratory tract infection (RTI), and conjunctivitis were considered risk factors only if they preceded or were concurrent with the first AOM episode for both AOM and ROM. Kappa statistics were used to determine interobserver agreement on OM diagnosis and comparison of algorithm with clinician diagnosis.23 χ2 test for linear trend was used to determine if AOM and ROM risk increased as the number of children ≤2 years to whom the infant was regularly exposed increased.23 Kaplan-Meier product limit survival curves were compared for time to AOM onset among infants with and without RTI, and distributions were compared by using the log rank χ2.22 Proportions and 95% confidence intervals (CIs) of exclusive breastfeeding by type of day care, and proportion of children attending day care were plotted against infant age.
A total of 611 individual women participated (11 withdrew during pregnancy, and 10 enrolled from the Diana cohort and during a subsequent pregnancy). This included 264 (78%) of 339 eligible Diana subjects, and 368 (93%) of 395 eligible responders from the other HMO enrollees. They delivered 627 infants (10 sets of siblings and 6 sets of twins). One infant with cleft palate and 1 with microcephaly were not eligible, and 29 withdrew before completing 6 months of follow-up, leaving 596 infants with at least 6 months of follow-up. Missing questionnaire and medical record abstract data varied by item, and ranged from 0% to 8%.
Among maternal participants, 96% were white, mean age was 31.3 years (SD = 3.6 years); median annual household income was $40 000 to $60 000; and 62% had a college degree. Fifty percent of the infants were female, and 56% had siblings. At age 2 months, 24% of all infants attended day care, 47% were exclusively breastfed, and 12% had 1 or more parent who smoked. Women who withdrew before 6 months of infant follow-up (n = 22 with available data) were significantly younger than participants (29.3 years, SD = 3.9 years; P = .02), but reported smoking status was similar. Although dropouts tended to be more likely to be employed, unmarried, and have an income <$40 000, differences between groups were not statistically significant. Compared with the other infants, Diana infants were more likely have household incomes <$60 000 (P < .001), but were less likely to be exclusively breastfed at 2 months of age or to have a mother who reported being a smoker (P = .04, for both).
Of urine samples collected on 460 subjects, 133 (29%) had detectable cotinine levels (>1 ng/mL), 86 (19%) had CCR levels ≥ 40 ng/mg, and 50 (11%) had CCR levels > 100 ng/mg. Median CCR was nondetectable, with a maximum value of 15 375 ng/mg. Mean difference in cotinine levels (range, nondetectable to 188) for the duplicate samples assayed blindly was 1.56 ng/mg; both assays of the pair resulted in the same conclusion of exposed/nonexposed, using the ≥40 ng/mg cut point. Pearson's correlation coefficient for the paired samples was 0.98, P < .01.
Thirty-nine percent of infants experienced an episode of clinician-diagnosed AOM between birth and 6 months of age, 48% had either AOM or OME, and 20% had ROM (≥2 episodes of AOM or OME) during this period. Onset occurred in the first 2 months for 10%, between 2 and 4 months for 20%, and between 4 and 6 months for 18%. Number of OM episodes (AOM and OME combined) during the first 6 months of life is shown in Fig 1, mean number of ear examinations per child was 5.9 (SD = 2.9). Among 612 children with some follow-up between birth and 6 months, there were 79 episodes of OM per 100 child-years. Kappa statistics were 0.54 and 0.65, comparing the middle ear diagnoses (OM or normal) of an investigator (G.S.G.) and a validated research nurse otoscopist with the sample of HMO pediatricians and nurse practitioners. The κ statistic was 0.92, comparing physician diagnoses with the computer algorithm for 3078 ears.
Risk Factors and Early OM
Day care attendance and feeding method changed substantially between 2 weeks and 6 months of age (Fig 2). Infants who entered center care before age 6 months and infants not in day care had similar rates of exclusive breastfeeding at 2 weeks of age, whereas children cared for in a family setting were less likely to be exclusively breastfed (P < .01). However, age at entry was not related to type of day care. Several day care variables were significantly (P < .05) related to AOM by age 6 months, including entry at 2 months of age or younger, center care, attendance ≥ 30 hours per week, ≥5 children in the day care setting, and ≥2 children 2 years or younger. Relationships between exposure to children ≤ 2 years of age (siblings or children in day care) and AOM and ROM can be seen in Table 1. χ2 tests for linear trend were significant (P < .01) for each comparison with the exception of the number of siblings in day care. In Cox's regression models, day care attendance (yes/no), day care type, and number of siblings were significantly related to early AOM onset when controlling for number of hours, number of children, and number of children ≤2 years old. Therefore, day care attendance and number of siblings were used in the multivariate models to predict AOM and ROM.
In addition to the day care hypothesis, univariate analysis relevant to other hypotheses about early infancy showed that low birth weight, young gestational age, or urine cotinine levels indicative of exposure to passive smoke (≥40 or >100 ng/mg) were not related to increased risk of AOM or recurrent OM in the first 6 months of life (see Table 2). In a similar manner, the investigation of maternal intake of vitamins, zinc, iron, or alcohol during pregnancy was largely negative except for the finding that the highest quartile of maternal dietary vitamin C intake during the last trimester of pregnancy was significantly inversely related to ROM risk (RR = 0.6, 95% CI, 0.4,0.9). However, high intake of vitamin C supplements and total dietary plus supplementary vitamin C intake were not significantly related to either outcome.
Other infant characteristics and early exposures that significantly (P < .05) increased or decreased the risk of AOM with univariate analysis are depicted in Table 2. Factors from the neonatal and early infancy period that were not related to early AOM or ROM and do not appear in Table 2 were use of vitamin and mineral supplements or introduction of foods by age 3 months, sleeping in the same room with a child or adult, supine feeding position, obstructed nasolacrimal duct, birth < 37 weeks, and neonatal breathing assistance. Infant dietary intake of calories, protein, folate, magnesium, iron, zinc, selenium, vitamin A, vitamin E, retinol, and carotene at 1 and 3 months of age was not significantly related to early AOM or ROM, but interinfant variability in specific nutrient intake was small. Maternal demographic factors (age, education, income, and race) were unrelated to outcome. Prenatal factors showing no association with OM or ROM were amniocentesis, third trimester maternal symptoms (nausea, vomiting, diarrhea, headache, fever, and allergy symptoms), and medications (aspirin, ibuprofen, acetaminophen, and allergy medications).
Dietary vitamin C was the only variable from the prenatal period that showed a univariate association with OM. The Diana cohort had dietary data from the entire pregnancy, affording us the opportunity to investigate trimester-specific relationships between vitamin C intake and early OM. In this cohort, high intake of dietary vitamin C during the third, but not the first and second, trimesters was associated with decreased risk of AOM (ARR 0.4; 95% CI, 0.3,0.6) and ROM (ARR 0.3; 95% CI, 0.2,0.5).24 Factors adjusted in the multivariate model were RTI, conjunctivitis, sibling OM history, prenatal maternal liquor intake, and day care. High vitamin C intake in the first trimester also had a borderline protective effect on ROM (RR = 0.50, P = .08, highest versus lowest 3 quartiles). Although intakes of dietary vitamin C in the first and third trimesters were correlated (r = 0.45), the relationship was weaker than correlations between intake in the first and second (r = 0.57) and second and third (r = 0.71) trimesters. Supplementary vitamin C and total vitamin C (diet plus supplement) during any trimester were not related to AOM or ROM. Intake of calories, folate, β carotene, and magnesium were correlated with vitamin C intake (r > 0.5), but none of these were related to early AOM or ROM.
Cox's proportional hazards models were used to predict early AOM and ROM (Table 3). Models revealed that RTI was the most important predictor for early AOM (ARR 7.1; 95% CI, 5.2,9.8), followed by day care attendance, paternal history of OM, maternal OM history, and having more than 1 sibling. Sibling OM history was of borderline significance at 1.4 (95% CI, 1.0,2.0;P = .09) in the full model and was significant in the stepwise reduced model (RR 1.7; 95% CI, 1.2,2.2). Having more than 1 smoking parent was of borderline significance in the full model (ARR 1.3, 95% CI, 1.0,1.8) but did not enter the stepwise model. For the most part, ARRs were the same or slightly lower than RRs in the forward stepping model. RTI, day care attendance, maternal OM history, conjunctivitis, and birth in the fall were significantly related to ROM in the multivariate model, but paternal and sibling OM history were not related (see Table 3). ARR of ROM in the first 6 months of life was nearly 9.5-fold for infants who experienced RTI.
Kaplan-Meier product-limit survival curves (Fig 3) revealed that 50% of infants with RTI and 25% without RTI had AOM by 6 months of age (log rank χ2, 36.83; P < .01). The proportional hazards assumption of the Cox regression was not met for RTI, but the violation was minor, occurring only between birth and 3 weeks of age when few AOM episodes were diagnosed. The proportional hazards assumption is that individuals with and without risk factors have proportional risks of the event, and these risks are constant during the time period studied. Therefore, the results from the Cox regression were considered acceptable.
Our results demonstrate that infectious diseases (respiratory infection, conjunctivitis), day care attendance, and other variables measuring likely exposure to pathogens are important determinants of AOM onset and recurrence in early life. Although bacterial agents are the predominant cause of AOM, viruses alone or in combination with bacteria have also been implicated.25 In addition, mechanisms by which viruses could predispose to bacterial invasion include damage to the middle ear and eustachian tube epithelium, decrease in mucociliary action, increase in bacterial adhesion to cells, impairment of cellular immune function and phagocytosis, and tubal dysfunction and negative middle ear pressure.25 Most infants in a study by Faden et al26 were colonized with OM pathogens (S pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis) by age 6 months. Colonization was more frequent during episodes of upper respiratory illness, but otitis-prone children had high rates of carriage even when well.27Furthermore, infants who were colonized in the first 3 months of life were twice as likely to have AOM by 6 months of age than noncolonized infants.27 Moreover, contact with infants and young children, particularly in day care settings, increases exposure to agents implicated in AOM. Dagan et al28 have demonstrated that children in day care settings with >6 children have significantly higher rates of S pneumoniae colonization than children not in day care or in smaller settings. Thus, an infant who enters day care early in life is at risk for colonization with pathogens, respiratory infection, and early AOM onset.
In the current study, many risk factors for early AOM and ROM were the same, suggesting they confer a common risk for single and multiple episodes. However, conjunctivitis and birth in the fall were significantly associated with OM recurrence but not with early AOM onset. Conjunctivitis and AOM share common pathogens (H influenzae and S pneumoniae) and risk factors (young age and day care).29 Indeed, the conjunctivitis-otitis syndrome is well known.29–31Infants in the current study who had low cord blood antibodies to twoS pneumoniae serotypes were more likely to have an early episode of conjunctivitis than infants with higher antibody levels.32 Conjunctivitis may be the source of pathogens for concomitant or subsequent AOM, a indicator of ongoing exposure to pathogens, or a marker of low antibodies and decreased resistance to OM-specific pathogens, all which could increase the likelihood of recurrent OM in the first 6 months of life. Infants born in the fall experience increased opportunity for exposure to upper respiratory infections because their early lives take place during winter and spring, peak seasons for respiratory disease. During the first 6 months of life, infant eustachian tube anatomy, immature immune systems, and declining passive antibody levels combine forces to predispose them to AOM. Also during this susceptible period, infants frequently enter day care, virtually assuring their exposure to AOM viral and bacterial pathogens.
Although well-controlled, prospective studies have reported a greater risk of ROM and chronic OME in children with a family history of middle ear disease,1 ,2 33–35 few researchers have examined the role of family history in early AOM onset. Teele et al2reported that having a sibling with an OM history increased the risk of both AOM and ROM in the first year of life, but parent history was not significantly related to either outcome, and maternal and paternal histories were not examined separately. Other researchers studying infants in the first year of life either have not investigated the role of family history35–39 or did not find a relationship between parent history and infant OM risk.40
Other investigators have reported that day care attendance by the infant or a sibling,36 ,41 ,42 having older siblings,41 respiratory infection,41 shorter duration of breastfeeding,41 ,43 and supine feeding36 were related to OM incidence or duration in the first 6 to 7 months of life. In the current study, only 17% of women breastfed exclusively for 6 months, which was associated with a decreased risk of OM in the univariate analysis, but was not significant in the adjusted multivariate models. Our study showed that infant but not sibling day care attendance increased AOM or ROM risk (see Table 1). Studies reporting an increased risk associated with sibling day care attendance were conducted in Sweden41 and Norway42 where children enter day care in the second year of life,42 not in the first 6 months of life, as in our study. Risk associated with a sibling in day care would be much lower than risk associated with infants' own day care exposures.
We used maternal report of parents' smoking status and urine CCR ratios to assess infant exposure to cigarette smoke. Having more than 1 smoker in the household showed a univariate relationship with AOM that was not significant in the multivariate model. Neither was the biomarker CCR related to early AOM onset. Other OM studies in the first year of life did not evaluate parental smoking,41 ,44 found no relationship between smoking and age at onset,36 ,37 ,42or found a relationship only for mothers who smoked heavily.45 The effect of passive smoke is probably cumulative and dose dependent, so that low doses and short duration of exposure were unlikely to effect changes in the middle ear and eustachian tube that would alter an infant's OM susceptibility by 6 months of age.
Karma et al44 conducted a retrospective investigation of prenatal conditions and AOM onset by age 3 months. Rates of maternal illness during pregnancy were similar in 96 cases and controls, but prematurity and asphyxia were both significantly more common in case infants with early AOM. Preterm birth was not associated with early AOM in the current study, but the number of preterm births was too small (n = 44) to provide adequate power to examine this hypothesis.
Among the prenatal exposures we evaluated as part of our a priori hypotheses, only dietary vitamin C intake showed a univariate relationship with early ROM. Vitamin C is a plausible risk factor because of its roles in immune function and synthesis of collagen,46 which is one of the structural components of the eustachian tube. Vitamin C intake, along with maternal alcohol intake during pregnancy, was significantly related to early AOM and ROM with both univariate and multivariate analysis in the cohort of Diana women.24 In the full cohort, however, women's dietary vitamin C was not significantly related to early AOM or ROM with multivariate analysis. The association between vitamin C and OM in the Diana subgroup but not the entire cohort is not likely to be caused by bias, because dietary data were collected before the infant's birth and physicians were unaware of the study hypothesis. Confounding is possible, because use of day care and sibling OM history, factors associated with the risk of infant OM in this cohort, were less common among mothers with high vitamin C intake.32 Risk factors that varied significantly between cohorts were unrelated to outcome, although unmeasured differences between the cohorts might explain the relationship between vitamin C and OM only in the Diana participants. One other study investigating vitamin intake and OM in the Northern Mariana Islands reported that baseline serum retinol and carotenoid levels did not predict subsequent OM in 3- to 5-year-olds.47
This study had several strengths. It was prospective, and with a 5% withdrawal rate, the likelihood of bias was reduced. Although there were many examiners, agreement on OM diagnosis between validated otoscopists and study physicians was good and physician ear examination findings had a very high level of agreement with a predetermined OM diagnostic algorithm. Multivariate analyses were done in which time-variable exposures (eg, RTI) were considered risk factors only if they occurred before the first AOM episode. Season of follow-up during the first 6 months of life, an important consideration in early AOM onset, was controlled by using season of birth as a covariate in the multivariate model. A weakness of the Cox model was that time of diagnosis was used as a surrogate for time of AOM onset because time of onset was unknown. However, in the case of infant AOM, time of diagnosis probably closely approximates time of onset. Participants were members of an HMO, which may not be representative of the larger population. Compared with a random sample of Minnesota mothers who participated in a survey about OM and risk factors, women in the current study were more educated and less likely to smoke, but had similar rates of breastfeeding and use of day care.48 Despite the fact that passive rather than active surveillance was used, participants had an average of six ear examinations by age 6 months. Thus, unrestricted access to care in an HMO resulted in frequent examination that would be comparable with active surveillance. Mothers who already had an affected child may have been more likely to enroll, which could overrepresent them in this study, but does not in itself bias the relationship between sibling OM history and early AOM onset in a subsequent infant. However, if physicians knew about an affected older sibling and were more likely to diagnose AOM in the infant when the TM changes were not definitive, the relationship between sibling OM history and early AOM onset could have been overestimated. In a similar manner, physicians are well aware of the relationship between respiratory infection and AOM, so the possibility that examiners were more likely to diagnose AOM in the presence of RTI cannot be ruled out.
In summary, both environmental and innate factors play an important role in early AOM onset. Exposure to siblings and other young children in day care and early respiratory infection predispose infants to AOM in the first 6 months of life, and sibling and parental history, probably surrogates for genetic factors, also significantly increase early AOM onset. Prenatal factors were not linked to early AOM onset with multivariate analysis. Later entry into day care, use of smaller day care settings, and improved handwashing and disinfection in these settings could decrease spread of respiratory pathogens and reduce early AOM onset. Consequently, reducing early AOM onset might result in declining rates of ROM and chronic OME, conditions responsible for a disproportionate number of physician visits, antibiotic treatments, and surgical procedures.
This work was supported in part by NIH Grants R01-DC01242 and P01-DC00133 from the National Institute on Deafness and Other Communication Disorders, the Minnesota Medical Foundation, and the Deafness Research Foundation.
We thank Devra Spindler, Ruth Selvius, Cynthia Roh, JoAnn Knox, and Kirsten Hase for participant follow-up, data collection, editing, and coding. We also thank Gerrie Barosso, project manager for the Diana study, for technical assistance in coding nutritional supplement constituents, and the Diana study staff for editing and coding food frequency questionnaires, file preparation, and transfer. The study would not have been possible without the HealthPartners research staff, the clinic physicians, nurse practitioners, and the nursing and laboratory staffs.
- Received August 31, 1998.
- Accepted December 29, 1998.
Reprint requests to (K.A.D.) Otitis Media Research Center, Box 396 UMHC, 420 Delaware St SE, Minneapolis, MN 55455.
- John EM,
- Savitz DA,
- Sandler DP
- Severson RK,
- Buckley JD,
- Woods WG,
- Benjamin D,
- Robison LL
- ↵Schappert SM. Office Visits for Otitis Media: United States 1975–90. Vital and Health Statistics, No. 214. Hyattsville, MD: National Center for Health Statistics; 1992;1–20
- ↵SPSS. Statistical Data Analysis. Chicago, IL: SPSS Inc; 1990
- ↵Lee ET. Statistical Methods for Survival Data Analysis. Belmont, CA: Lifetime Learning Publications; 1992
- ↵Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed. New York: John Wiley & Sons; 1981:211–236
- ↵Daly KA, Meland M, Brown J, Lindgren B, Westover D, Giebink GS. Epidemiology of early otitis media. In: Lim D, Bluestone C, Casselbrant M, Klein J, Ogra P, eds. Recent Advances in Otitis Media. Proceedings of the 6th International Symposium. Hamilton, Ontario, Canada: BC Decker Inc; 1996:33–35
- Dagan R,
- Melamed R,
- Muallem M,
- Piglansky L,
- Yagupsky P
- Groothuis JR,
- Thompson J,
- Wright PF
- Bodor FF,
- Marchant CD,
- Shurin PA,
- Barenkamp SJ
- ↵Daly KA, Brown JE, Lindgren BR, Meland MH, Giebink GS. Relationships between otitis media risk factors in early life. In: Tos M, Thomsen J, eds. Otitis Media Today. Proceedings of the 3rd Extraordinary Symposium on Recent Advances in Otitis Media. The Hague, The Netherlands: Kugler Publications; 1998:1–5. In press
- Owen MJ,
- Baldwin CD,
- Swank PR,
- Pannu AK,
- Johnson DL,
- Howie VM
- Paradise JL,
- Rockette HE,
- Colborn K,
- Bernard BS,
- Smith CG,
- Kurs-Lasky M
- Kværner KJ,
- Nafstad P,
- Hagen JA,
- Mair IWS,
- Jaakkola JJK
- Duncan B,
- Ey J,
- Holberg CJ,
- Wright AL,
- Martinez FD,
- Taussig LM
- Ey JL,
- Holberg CJ,
- Aldous MB,
- et al.
- Daly KA,
- Selvius RE,
- Lindgren B
- Copyright © 1999 American Academy of Pediatrics