Window of Susceptibility to Acute Otitis Media Infection

As previously described,¹²  children were prospectively enrolled at about 6 months of age and followed to 60 months of age from June 2006 to August 2017. The children were all from a primary care pediatric practice in Rochester, NY comprised of a typical mixed demographic of largely middle class, health care-insured families but including both wealthy and below poverty line families on public assistance, and broadly representative of the racial and ethnic diversity in the community. The diagnosis of AOM was made based on the AAP guidance wherein a presumed middle ear effusion and a full or bulging tympanic membrane were required.¹⁴ 

Almost all episodes (estimate >85%) of clinically diagnosed AOM were confirmed by culture of middle ear fluid collected by tympanocentesis up to 36 months of age. Children experiencing 3 AOM episodes within a 6 month time span or 4 episodes within a 12-month time span, virtually all confirmed by tympanocentesis, were termed otitis prone.^2,14  Multiple risk factors known to impact the occurrence of AOM were recorded. The electronic medical record of all children was accessed for details of demographic and epidemiologic risk, vaccinations, AOM diagnoses from birth to child age 5 years old. We have previously reported that AOM is infrequent after 36 months old in our study population¹⁷  and data from those children were so sparse that they were not included in the current study. Infrequently AOM diagnosis was made at Urgent Care centers and when that occurred, typically (estimate >85%) parents brought the child to the study practice and AOM clinical diagnosis was confirmed the following day by reexamination by 1 of the study clinicians and tympanocentesis performed. The details of the methods and culture results from the middle ear fluid from the study children have been reported previously.^1,18,19  Tympanocentesis cultures collected from the study cohort yields Streptococcus pneumoniae, Haemophilus influenzae or Moraxella catarrhalis by culture or polymerase chain reaction in ∼90% of middle ear fluid (MEF) sample.¹  The Rochester General Hospital IRB approved the study and written informed consent was obtained from parents before enrollment in the study.

To explore the WOS concept, we needed to define more precisely what the WOS is. When AOMs are spaced within a relatively short period of time, this has been used in the past to define the otitis prone child.¹⁴  Here we evaluated the occurrence of AOMs over time in individual children to ascertain the best definition that fit the data; a cumulative distribution curve of AOM occurrences in the cohort identified patterns of AOMs closely spaced in time and gaps between consecutive AOMs. The data identified AOMs spaced more than 6 months apart as separate AOM infections or a separate WOS. Hence, our definition of WOS in this report is an AOM sequence of 2 or more with no gap longer than 6 months. A gap of 6 months followed by 1or more AOMs is an isolated event or a different WOS for an individual child.

Analysis of demographic and risk factor covariates was based on the 2-sample test of Poisson rates. For the prediction of the total number of AOMs, we used a quasi-poisson generalized linear model. Mild under-dispersion (variance smaller than expected) occurred in a Poisson model,²⁰  likely caused by some correlation among children, especially those attending the same daycare, but we do not have specific data that would allow us to include this in the model. Natural logarithm as the link function was used. The model fit was confirmed by plots of a given functional relationship with respect to a continuous predictor, such as the age at AOM 1 through AOM 4, respectively, and the averages in a +/− 1-month window around the age specified. The distributional assumptions of the models were verified with residual analysis. The reported P values for the significance of predictors were based on the t-statistics for the discussed predictor in the model. Based on the quasi-poisson generalized linear model, a pseudo-R² was determined based on the method of Cohen.²¹ 

Analysis of demographic and risk factor covariates: sex, race and ethnicity, daycare attendance, breastfeeding siblings in the home, smoking in the home, atopy, and family history of otitis media, identified daycare attendance as a significant factor (P < .001) associated with number of AOMs in the child population (Table 1). Children attending daycare had an average of 2.7 AOMs, whereas those not attending had an average of 2.0 AOMs.

AOMs were infrequent among children before 6 months old (n = 23 [8%] of 286, see Supplemental Table 3), but then AOMs increased sharply and had a broad peak from child age 6 to 21 months old, accounting for 80% of AOMs, when they decreased in frequency through the remainder of the observation time interval. The number of AOMs per child per 2-month period is shown in Supplemental Fig 6.

Figure 1 shows the cumulative distribution function of the gaps between sequential AOMs. The line in Figure 1 becomes less steep beyond the threshold of 6 months, suggesting that sequential AOMs spaced more than 6 months apart should not be included within the same WOS. The rate of change in the fraction of AOMs as shown in Figure 1 versus the gap between AOMs is illustrated in Supplemental Fig 7, further illustrating that the rate of change is high up until the 6-month gap threshold.

Figure 2A shows 67 children with 2 AOMs. Of those, 49 had closely spaced AOMs that was a WOS with the mean length of 1.9 month (95% confidence interval [CI] [1.5–2.3]). In the remaining 18 children, isolated cases of AOM occurred with gaps wider than 6 months. Figure 2B shows 41 children with 3 AOMs. Of those, 38 children had closely spaced AOMs that was a WOS with the mean length of 3.6 months (CI 2.9–4.4). In the remaining 3 children, isolated cases of AOM occurred with gaps wider than 6 months. Figure 2C shows 28 children with 4 AOMs. All of them had at least 1 WOS, and 5 of them had 2 WOS. The mean WOS was 3.9 (CI 3.0–4.7). Figure 2D shows children with more than 4 AOMs. The mean WOS among the 35 children with more than 4 AOMs was 7.0 CI 6.0–8.0). Of note, 4 children in Fig 2D had 2 WOS periods separated by more than 6 months without any intercurrent AOMs. For 8 out of 9 children who had 2 episodes of WOS (Fig 2 C and D), we found that the child changed rooms in the day care setting between their separated WOS times, moving from the infant room to the toddler room. Details are provided in Supplemental Table 4.

Figure 3 shows the overall cumulative distribution of WOS length. In our cohort, 72% of the WOS intervals were shorter than 5 months and 97% were shorter than 10 months.

We have published previously on risk factors associated with recurrent AOM in this child cohort²² ; however, those analyses used the- specific definition of the AAP and American Academy of Otolaryngology Head and Neck Surgery of 3 AOM within 6 months or 4 AOM within 12 months. Here we undertook a novel analysis of risk factors associated with 2, 3, 4, or 5 AOMs in a newly defined WOS terminology. We explored the factors shown in Table 1 and sought to predict the expected total number of AOMs in the first 3 years of the child’s life. The only significant predictors were the child’s age at the first AOM (age at AOM1) (P < .001) and daycare attendance (P = .03), R² = 0.3. Different models were suitable for the children who attended daycare versus those who did not. Figure 4A shows the model for children at the time of their first AOM who attended daycare. There was more variability when age was older than 15 months because of fewer observations in that range. Figure 4B shows the model for children who did not attend daycare. For children who did not attend daycare, the model predicts that only approximately 1 additional AOM after the first one happened at any age. Figure 4 C, D, and E shows a model for children who have experienced a second, third, and fourth AOM, respectively. For second, third, and fourth AOMs, the model was not influenced by daycare. Details for Fig 4 B, C, D and E are in Supplemental Information (Supplemental Figs 8–12, Supplemental Tables 5–9). Supplemental text provides the formula for input of child age, and AOM1, AOM2, AOM3, or AOM4, and daycare to derive a prediction for individual children regarding likelihood of future AOM frequency.

Table 2 provides finite predictions for future AOMs. For example, for children who experienced their first AOM before the age of 6 months, as many as 2.7 additional AOMs on average (with a 95% CI [2.2–3.1]), can be expected. On the other hand, when the first AOM occurs after 15 months of age, only up to 1 additional AOM, on average, can be expected. For children who experienced their second AOM before the age of 6 months, as many as 2.9 additional AOMs, on average (with a 95% CI [2.2–3.6]), can be expected.

In the study cohort, all of whom had at least 1 AOM, 69 (23.8%) children met the commonly employed definition of otitis prone of 3 AOMs within 6 months or 4 AOMs within 12 months window.¹⁴  Among those attending or not attending day care or daycare status not reported (only 12 cases), 23.9%, 20.7% and 50.0%, respectively, met the definition. Parents of 31 children requested and had their child receive pressure equalizing tympanostomy tubes (PETs) after experiencing 3 AOMs (n = 15) or >3 (n = 16). After insertion of PETs, 16 (52%) children had no additional AOMs, 9 (29%) children had 1 AOM for a total of 81% (Fig 5). Three (10%) children had 2 AOMs and 3 (10%) children had 3 AOMs after tubes were placed and they tended to have had recurrent AOM at a younger age, consistent with results described above. For 10 (67%) out of 15 children that had any AOMs, they occurred within 6 months of PET insertion and for the remaining 5 (33%) children, an AOM occurred 10 to 16 months later (Supplemental Fig 13).

In the past decade major changes have occurred in the incidence of AOM and recurrent AOM caused by updated recommendations on diagnostic criteria and treatment¹⁴  and introduction of pneumococcal conjugate vaccines. Here, for the first time, we report recurrent AOM incidence with respect to age of occurrence, the influence of day care attendance, and other risk factors in individual children to identify the length in time of the WOS to AOM in early life. We developed prediction models to assist clinicians in understanding and explaining to parents what benefit to expect if tympanostomy tubes are inserted based on the child’s age, number of AOMs, and daycare attendance. Thirdly, we assessed what happens to children after tubes are inserted.

AOMs are infrequent among children before 6 months old but when they occur and the child is in daycare, our results suggest that the child is destined to recurrent AOM with 3 to 4 future episodes yet to come. This possibility should be considered in the context of possible adverse effects of systemic antibiotics on vaccine-induced antibody levels,⁵  contribution to antibiotic resistance from exposure,^23,24  and side effects from antibiotic consumption.

AOMs are most frequent in children from 6 to 14 months of age at the rate of at least 0.3 AOMs per 2-month interval, but then the occurrence of AOM is significantly reduced (Supplemental Fig 6). We found that the WOS of AOM is very short for most children and depending on the age of the child at that moment and if in daycare or not, the likelihood of further future AOMs can be reasonably predicted with good accuracy. Of course, there will be exceptions, and we provide the percentages to help the clinician set expectations. Importantly, the widely held notion that the child with 2, 3, or 4 AOMs in closely spaced time is destined to experience at least the same number going forward needs to be reassessed based on the findings of our report. Faced with such children, following current AAP and American Academy of Otolaryngology Head and Neck Surgery guidelines, many receive tubes and the cessation of AOM is attributed to the surgery when, in fact, our data suggests the WOS may have ended and the child would have experienced an end to AOMs with patience and watchful waiting.

Children in daycare were identified as at higher risk for recurrent AOM, consistent with prior reports.^1,25  The quantitative difference in likelihood of recurrent AOM among children attending daycare versus those not in daycare was clinically and statistically significant (Table 2). Mingling with a new group of children and greater freedom to move about may explain exposure to new pathogens that triggers second episodes of WOS in children who attend daycare. For children not in daycare, occurrence of AOM is not only less frequent but rarely is followed by recurrent AOM sufficient in number to lead to a recommendation for tympanostomy tubes.

Clinicians dealing with children who experience recurrent AOM, especially when AOMs occur in rapid succession in a short time interval, would benefit from data to inform their decisions and discussions with parents regarding options of management. The predictive models described in this report should prove useful in that regard. Using input of child age, daycare attendance, and AOM number in a sequence of AOMs, tabulation and simple mathematical formulas are provided to allow clinicians to be more precise in predicting likely further numbers of AOMs (see imbedded macro in Supplemental Information of this report).

After tympanostomy tubes are inserted, some children experience no further AOMs and others subsequently have additional AOMs manifest as otorrhea from the tube. If the child has experienced 3 or 4 AOMs in rapid succession, the expectation is that 3 or 4 more AOMs are on the horizon. In our cohort, half the children had no further AOMs and 81% had none or 1. If AOM occurs frequently after tube surgery, our data suggests that most likely (83%) the child had at least 3 AOMs before 9 months of age.

Our study has limitations. The study cohort was derived from a single community in upstate New York, and approximately three-fourths of the children were reported as non-Hispanic white. To confirm the clinical diagnosis of AOM, the children had tympanocentesis completed, which is not standard practice.

The performance of tympanocentesis to confirm the clinical diagnosis of AOM is a strength since it removes diagnostic uncertainty in a setting where AOM may be over-diagnosed. Drainage of MEF using tympanocentesis may have a therapeutic benefit, as previously reported.²⁶  The children who underwent tympanocentesis did not exhibit a different pattern in our models predicting the total number of AOMs based on the age at which the first, second, third, or fourth AOM was observed. Also, Figure 5 shows that tympanocentesis at an early age (thus coinciding with early AOMs) tends to be followed by more AOMs versus tympanocentesis at later age. These strengths and limitations of our study should be considered by clinicians in interpreting our results.

In conclusion, recurrent AOM occurs in a narrow WOS and number of AOMs can be predicted at time of AOM based on child age and daycare attendance. Insertion of tympanostomy tubes likely occurs in many children after the WOS to recurrent AOM has passed or only 1 more AOM may be prevented at most. Multiple courses of antibiotics have consequences as do tympanostomy tubes^2,27,28  and assessment of risk and benefit should be individualized.

We thank the physicians, nurses, and staff of Legacy Pediatrics, the parents who consented, and the children who participated in this long and challenging study.

AAP

American Academy of Pediatrics

AOM

acute otitis media

WOS

window of susceptibility