1. Introduction

Concepts regarding the diagnosis and management of acute otitis media (AOM) have evolved as our understanding of the adverse impact of excessive antibiotic usage on the spread of resistant bacterial pathogens increases [1]. Excessive antibiotic treatment also alters nasopharyngeal colonization predisposing to a subsequent episode, which may lead to recurrent AOM [2]. Antibiotic induced alterations to the microbiome in early childhood appears to predispose to childhood obesity [3, 4]. Young children during their second year of life have the highest AOM rates due to anatomic and immunologic reasons and children between the ages of 1 and 3 years account for over 40% of otitis related expenditures in children under age 14 [5]. Therefore interventions to limit excessive AOM related antibiotic use in this young, high risk population need further investigation. Two possible approaches to reducing excessive antibiotic use are first, shared decision making that provides parents an option for a safety net antibiotic prescription (SNAP) and second, anticipatory ear pain counseling to reduce parent initiated primary care (PC) and urgent care emergency department (UC/ED) visits.. Both of these approaches have the potential to reduce costs associated with unnecessary AOM visits especially in UC/ED settings as well as excessive antibiotic use.

Shared decision making provides parents with unbiased information on the risks and benefits of delaying antibiotic treatment for uncomplicated AOM episodes. While small studies suggest it is acceptable to parents and is viewed as beneficial, no large-scale studies have documented its impact [6-10]. Anticipatory ear pain counseling at 12 and 15 month routine preventive care visits can address home ear pain management including when and where to seek care. Watchful waiting for ear pain associated with uncomplicated AOM might reduce overall AOM visits, and UC/ED visits and therefore reduce antibiotics as well as costs. Ear pain counseling implemented in Rochester, Minnesota private practices reduced AOM visits in both settings but the unit of randomization was the practice rather than the patient [11]. We know of no studies that randomized at the child level, enrolled low income children on public health plans (Medicaid and Child Health Plan), or included clinic rather than private practice settings. We therefore undertook this study with the following aims:

1. Determine if a program of anticipatory ear pain counseling at the 12-15 month routine preventive care visit with largely Hispanic children enrolled in Medicaid reduces overall, PCC, and UC/ED AOM visits during the following year.
2. Determine whether socio-demographic characteristics affect the impact of anticipatory ear pain counseling on AOM visit incidence rates.

2. Materials and Methods

2.1 Design

This was a randomized controlled trial of ear pain anticipatory counseling that began in 2009 and ended 2 years later in the Child Health Clinic at Children's Hospital Colorado, a large, metropolitan, hospital-based teaching clinic serving low-income families. Children assigned to the active arm in the current study received anticipatory ear pain counseling. Counseling related to an early childhood language development program was used as the active control. The study was approved by the Colorado Multiple Institutional Review Board.

2.2 Participants

We enrolled children aged 12-15 months seen at the clinic for their routine preventive care visit in the study. Children were excluded if they had ventilating tubes or any condition that predisposed them to recurrent AOM such as Down’s syndrome, cardiac disease, or an immunodeficiency disorder. Children were also excluded if their family's primary language was something other than English or Spanish. The research assistant approached eligible families during the child's office visit and obtained written informed consent from all agreeable participants. Since the research assistant was not present during all clinic hours, this was a convenience sample. Data on the mothers who refused to participate were not kept but the total number refusing was less than 10. Each family was then assigned using a computer generated randomization method to receive either the active ear pain anticipatory counseling program or the active control language development program, which were completed during the visit.

2.3 Counseling intervention

The research assistant provided a structured 10 minute counseling session using 10 PowerPoint slides specific to the subject's assigned group. This was done in the exam room during the child's clinic visit using a portable laptop computer to display the slides, and a copy of the slides was provided to each family at the end of the session. The ear pain counseling material was based on the materials used in the McWilliams study, which were provided to us. The material   reviewed concepts such as how to recognize ear pain and safely provide pain relief, and how to recognize danger signs that require urgent medical attention. Families were also encouraged to schedule an appointment in the PCC for a possible ear infection rather than going to the UC/ED facility. The research assistant provided and reviewed proper dosing instructions for acetaminophen and ibuprofen, and provided a prescription for antipyrine/benzocaine analgesic ear drops to each family to use as pain relief if their child did develop ear pain in the subsequent 12 months.

The language development materials explained the importance of frequent conversations between parents and children and of using encouraging rather than discouraging comments. The research assistant reviewed the importance of daily reading using a provided children's book as an example.

2.4 Data collection and outcome measures

All participating families completed a demographics survey that included information on race/ethnicity, insurance status, language(s) spoken in the home (preferred language for the intervention material), and family composition. In addition, participants completed the StimQ, the Parent Evaluation of Developmental Status, and the MacArthur-Bates Communicative Development Inventory. The EPIC electronic medical records system was used to determine the number and site of visits for AOM, whether antibiotics were prescribed at the visit, and the number of antibiotic courses prescribed in the intervention and control groups. We counted visits if there was a primary or secondary International Classification of Diseases, Ninth Revision (ICD-9), diagnosis reflective of ear pain or otitis, including ICD-9 codes 380.10, 380.22, 381.00, 381.01, 381.10, 381.4, 382.00, 382.01, 382.3, 382.4, 382.9, and 388.70. The electronic medical record for each child was reviewed for a diagnosis of AOM using all otitis related diagnostic codes for 12 months following their initial index visit. We did not review charts to validate clinical findings that met predefined AOM criteria.

The site of each subsequent AOM related visit was documented (primary care clinic, emergency department, urgent care clinic, otolaryngology clinic) within the 6 metropolitan Denver Children’s Hospital Colorado network of care sites as well as the antibiotic that was prescribed. The electronic medical record review documented the number of AOM related visits for each child in total and by site of care as well as the time from the initial index visit to the first AOM related visit with an associated antibiotic prescription. The data did not include any AOM visits outside the network of care sites. We did not determine if any of the subjects no longer used primary care services at Children’s Hospital Colorado.

2.5 Statistical analysis

We used the non-parametric Wilcoxon signed-rank test used to test for difference between groups for time to first AOM comparisons. Interactions between the study group and dichotomous demographic variables were tested in negative-binomial regression models. When socio-demographic data was missing, the enrollee was not included in that sub-analysis. Final stratified regression models represent the relationship between ear pain visits, the study group, and the dichotomous demographic characteristic. Each stratified model was analyzed to represent the appropriate intervention vs. control comparison for each level of the demographic factor [e.g., Post High School education (intervention vs. control) and ≤ High School education (intervention vs. control)].

3. Results

3.1 Study population

The socio-demographic characteristics of the total population studied and the cohorts that received and did not receive ear pain counseling are shown in Table 1. Each cohort included 155 children. Half of the families were Hispanic. Slightly over two thirds (67.2%) were either married or living as couples. Approximately one quarter of the mothers (22.8%) did not graduate high school or have a GED, 43.3% had some post high school education and 33.9% had a high school education. Nine of 10 children (89.9%) were enrolled in Medicaid. The ear pain counseling intervention cohort and control cohort were comparable with respect to all socio-demographic characteristics.

*represents the probability of a difference in the demographic characteristic and study group. Student's t-test used for mother's age at baseline. Chi-square test of proportions or Fisher’s exact test (for small cell sizes) was used for all other tests.

Table 1: Socio-demographic characteristics of the study cohort, by who received (Intervention) and did not receive (Control) ear pain counseling.

3.2 Distribution of children according to the number of AOM visits by setting and the proportion of children with at least 1 AOM visit

Table 2 shows the distribution of the number of children according to the number of AOM visits during the 12 months following study enrollment by PCC and UC/ED settings. Among the 310 children enrolled in the study and followed for 12 months, 30.6% had at least 1 AOM visit and 4.5% had recurrent AOM defined as 3 or more AOM visits. Table 3 displays the number and percent of children with at least 1 AOM visit during follow-up by study group according to sociodemographic characteristics. The observed higher proportion of Hispanic children with at least 1 AOM visit (33.5% 52/155) compared to Non-Hispanic African American and White children with at least 1 AOM visit (24% 25/106) approached significance ( p=0.08). The proportions of African American children (23%, 18/78) and White children (25%, 7/28) with at least 1 AOM visit were similar. The shorter time interval in days from entry to their first AOM visit for Hispanic children (mean 111.2, median 94) compared to Non-Hispanic White and African American children (mean 148.3, median 139) also approached significance (p=0.07).

3.3 AOM visit incidence rates

Table 4 displays the incidence rates of AOM visits per year, per 1,000 children, by setting and study group obtained by medical record review. In the total study population, the incidence was 490 AOM visit episodes per 1000 child years. The site of care for the first visit with AOM was almost equally distributed between the PCC (51.2%) and the UC/ED settings (48.8%). For all children, the PCC setting incidence rate (232) and the UC/ED setting rate (226) were similar. For all settings, Hispanic children had an incidence rate of 561 compared to 419 in Non-Hispanic children (p=0.18). Hispanic children had a higher PCC incidence rate (297) compared to Non-Hispanic children (168) (p=0.05). The rates in the UC/ED for Hispanics (219) and Non-Hispanics were similar (232) (p=0.85). Antibiotics were prescribed in 97% of the visits associated with an AOM diagnosis and were prescribed similarly in all settings for visits with an AOM diagnosis. We did not have data documenting the percentage of prescriptions that were filled.

CHC=Children's Health Clinic; ED/UC=emergency department or urgent care facility

Table 2: Distribution of the number of children to number of otitis related visits during the 12 months following study enrollment, by study site.

*Chi-square test of proportions or Fisher’s exact test (for small cell sizes) used to test if there is a difference between demographic characteristic and study group, among patients with ≥ 1 visits during study follow-up

Table 3: Number and percent of children with at least one Otitis visit (≥ 1) during follow-up, by study group.

*rate of Otitis visits per year, per 1000 children; **estimated relative risk from negative binomial regression testing the likelihood there is no difference in the number of visits and study group. Model included number of visits and study group; †all Otitis visits to an emergency department (ED)/urgent care facility (UC) resulted in an antibiotic prescription, therefore, only ED/UC total visits were modeled

Table 4: Rate of Otitis visits per year, per 1,000 children, by site and study group.

3.4 Ear pain anticipatory counseling outcomes

The ear pain anticipatory guidance counseling did not reduce the proportion of children with at least 1 AOM visit (counseling: 29.0%: 45/155; control: 32.3%: 50/155 p=0.54); the proportion of children with 3 or more AOM visits (counseling: 5.8%: 9/155; control: 3.2%: 5/155 p=0.41); or the total AOM visit incidence rates (counseling: 471; control: 510; p=0.71). Table 4 displays the yearly AOM visit incidence rates by PCC and UC/ED settings for the counseling and control cohorts. The intervention did not shift AOM visits from the UC/ED settings to the PCC setting. There were no significant differences in the proportion of families seeking care at their PCC site between the cohorts (counseling: 51.2% vs. control: 57.8%, p=0.53) or in the AOM incidence rates in the settings.

3.5 Socio-demographic risk characteristics analyses

Analyses were carried out to determine if sociodemographic characteristics of race/ethnicity (Hispanic, African American, Non-Hispanic White), mothers living status (single parent, 2 parent) and mother’s education (less than high school, high school/GED, Post high school) might influence the response to the ear pain anticipatory guidance. There were no significant differences in the proportion of children with at least 1 AOM visit when the cohorts were stratified by their socio-demographic characteristics (Table 3). However, a trend was identified as 20% (8/40) mothers with a high school education who received counseling compared to 36% (21/58) of mothers who did not receive counseling had at least 1 AOM visit (p=0.08). A regression analysis modeling the total number of AOM visits with interaction between the cohort, mother’s education and Hispanic ethnicity is shown in Table 5.

*rate of Otitis visits per year, per 1000 children; **estimated relative risk from negative binomial regression testing the likelihood there is no difference in the number of visits and study group. Model included number of visits, study group, education, ethnicity, and an interaction between education and ethnicity.

Table 5: Number of Otitis visits per total children in group, with rate and 95% confidence interval of Otitis visits per year, per 1000 children.

Among mothers with a high school education or less, the yearly incidence of AOM visits was less in the intervention group (361) compared to the control group (620). However the difference only approached significance (p=0.08). The effect of ear pain anticipatory guidance was in the opposite direction for mothers with greater than a high school education as the yearly AOM visit incidence was greater in the intervention group (520) compared to the control group (260) (p=0.08). There were no AOM incidence differences with and without counseling noted between Hispanic and Non-Hispanic families. It is unclear why mothers with more education who did not receive counseling should have very low AOM visit incidence rates compared to mothers with counseling.

4. Discussion

We studied AOM visits during the second year of life; the age shown to have the highest annual incidence rates in order to be consistent with the McWilliams study [11, 14]. The AOM incidence rate during the second year of life calculated from 209,180 children with commercial insurance included in the Truven Market Scan database for 2009 reported by Tong et al. was 516 AOM visits per 1000 child years (95% CI: 513.3-517.7) [13]. The mean incidence from 2008-2014 was 503.9 with a range of 479-521. Race/ethnicity data was not reported. Kaur et al. reported a slightly lower AOM incidence rate of 480 for 523 children followed closely in a Rochester, New York private pediatric practice during their second year of life [15]. The children enrolled in this prospective study were 68.9% Non-Hispanic White, 12% African American, 3.6% Hispanic, and other 15.4%. A large European prospective study reported by Liese et al., carried out from 2008-2010, followed 1113 children aged 0-2 years of age and documented slightly lower incidence rates: 392 (95% CI 342-447) for Spain, 344 (298-394) for Sweden, 311 (267-359) for Germany 247 (203-298) for the UK, and 193 (160-232 ) for Italy [16]. Our 490 AOM visit incidence rate for 310 children with Medicaid was almost midway between the Rochester study and the national Truven Market database study. However, our Hispanic rate of 561 was slightly higher than the rates from all of these cited studies. While nationally, approximately 50% of births in the United States are covered by Medicaid and Hispanic children account for an increasing proportion of these Medicaid births [12] there is limited data on the incidence of AOM for Hispanic children enrolled in Medicaid, after the introduction of pneumococcal vaccine [13]. Our study suggests this population may have a higher risk of these early infections.

Study limitations included our inability to review charts to validate clinical findings that met predefined AOM criteria, having a .convenience sample that could lead to bias dependent on those who elect to participate (only 10 mothers refused to participate), and our inability to capture AOM visits outside of the Children’s Hospital Colorado network of care system. However, the system is quite extensive with 6 UC/ED sites in the Denver metropolitan area. However, Hispanic families may be likely to stay in the system then Non-Hispanic families.

In our study 30.6% of children had at least 1 AOM visit during the year of monitoring and 4.5% had 3 or more infections. The higher rate for Hispanic children (34%) compared to African American (23%) and Non-Hispanic white (25%) children approached significance. Kaur reported that 23% of their population had at least 1 AOM episode during the first year of life, but he did not report the number of episodes during the second year of life.

There were substantial differences in the relative AOM visit incidence rates for PCC versus UC/ED settings in our population predominately on a public plan compared to children with commercial insurance. In our study, the site of care was almost equally distributed between the PCC (51.2%) and the UC/ED (48.8%) settings and the incidence of AOM visit episodes seen in the PCC setting was 232 compared to 226 in the UC/ED setting. In the Tong et al study of children with commercial insurance the incidence rate in the primary care outpatient setting (479.5) was much higher compared to the UC/ED setting (35.7) [14].

Several studies have documented that children enrolled in Medicaid have high rates of UC/ED utilization for non-acute care [17-20]. Improving federal qualified health center access in California did not appear to decrease ED visit rates for children enrolled in Medicaid [21]. The reasons for the high utilization of UC/ED settings by children with Medicaid are multifactorial. They include the organizational structure of the primary care setting (the degree of clinician continuity, the length of waiting times before seeing the clinician, the ease of same day appointment scheduling, and the availability of after-work hour appointments), parental work demands, accessibility, financial implications, and care seeking behaviors. Work demands limit the ability of working parents to leave work to take their children for care. Parents may choose to go to closer, more convenient UC/ED settings, especially if they are in the same hospital system. The lack of financial disincentives to UC/ED use may also impact decision-making. Finally, care-seeking behaviors may be influenced by cultural factors as well as past experiences.

We could not replicate the outcomes of an ear pain counseling program reported by McWilliams et al, who randomized at the practice level. Our study unlike the Mayo Clinic program enrolled a low income, predominantly Hispanic and African American population who were randomized at the patient level. Many of the factors that appear to influence the use of UC/ED settings by children enrolled in Medicaid may also influence their response to ear pain counseling.

5. Conclusion

Anticipatory ear pain counseling in a hospital based PCC serving children enrolled in public plans does not reduce PCC or UC/ED AOM visit incidence rates. The higher proportion of Hispanic children with at least 1 AOM visit compared to Non-Hispanic children approached significance and Hispanic children also tended to have a shorter interval from entry to their first AOM visit.

References

1. Bauchner H, Marchant CD, Bisbee A, et al. Effectiveness of Centers for Disease Control and Prevention recommendations for outcomes of acute otitis media. Pediatrics 117 (2006): 1009-1017.
2. Dagan R, Leibovitz E, Cheletz G, et al. Antibiotic treatment in acute otitis media promotes superinfection with resistant Streptococcus pneumonia carried before initiation of treatment. J Infect Dis 183 (2001): 880-886.
3. Saari A, Virta LJ, Sankilampi U, et al. Antibiotic exposure in infancy and risk of being overweight in the first 24 months of life. Pediatrics 135 (2015): 617-626
4. Scott FI, Horton DB, Mamtani R, et al. Administration of antibiotics to children before age 2 years increases risk for childhood obesity. Gastroenterology 151 (2016): 120-129.
5. Bondy J, Berman S, Glazner J, et al. Direct expenditures related to otitis media diagnoses: extrapolations from a pediatric Medicaid cohort. Pediatrics 105 (2000).
6. Taylor JA, Kwan-Gett TSC, McMahon Jr., EM. Effectiveness of an educational intervention in modifying parental attitudes about antibiotic usage in children. Pediatrics 111 (2003): 3548-3554.
7. Merenstein D, Diener-West M, Krist A, et al. An assessment of the shared-decision model in parents of children with acute otitis media. Pediatrics 116 (2005): 1267-1275.
8. Cates C. An evidence based approach to reducing antibiotic use in children with acute otitis media: controlled before and after study. Br Med J 318 (1999): 715-716.
9. Siegel RM, Kiely M, Bien JP, et al.. Treatment of otitis media with observation and a safety-net antibiotic prescription. Pediatrics 112 (2003): 527-531.
10. Marchetti F, Ronfani L, Nibali SC, et al. Delayed prescription may reduce the use of antibiotics for acute otitis media: a prospective observational study in primary care. Arch Pediatr Adolesc Med 159 (2005): 679-684.
11. McWilliams DB, Jacobson RM, Van Houten HK, et al. A program of anticipatory guidance for the prevention of emergency department visits for ear pain. Arch Pediatr Adolesc Med 162 (2008): 151-156.
12. National Vital Statistics Reports 62 (2013).
13. Byrns PJ, Bondy J, Glazner JE, et al. Utilization of services for otitis media by children enrolled in Medicaid. Archives of pediatrics & adolescent medicine 151 (1997): 407-413,
14. Tong S, Amand C, Kieffer A, et al. Trends in health care utilization and costs associated with acute otitis media in the United States during 2008-2014 BMC Health services research 18 (2018): 318.
15. Kaur R, Morris M, Pichichero ME. Epidemiology of Acute Otitis Media in the Post pneumococcal Conjugate Vaccine Era Pediatrics 140 (2017): 20170181.
16. Liese JG, Silfverdal SA, Giaquinto C, et al. Incidence and clinical presentationof acute otitis media in children aged, 6 years in European medical practices. Epidemiol. Infect 142 (2014): 1778-1788
17. Pukurdpol P, Wiler JL, Hsia RY, et al. Association of Medicare and Medicaid insurance with increasing primary care-treatable emergency department visits in the United States. Acad Emerg Med 21 (2014): 1135-1142
18. Rasooly IR, Mullins PM, AlpernER, et al. US Emergency department use by children,2001-2010 Pediatric Emergency Care 30 (2014): 602-607.
19. Brousseau D, Hoffmann R, Nattinger A, et al. Quality of primary care and subsequent pediatric emergency department utilization. Pediatrics 119 (2007): 1131-1138.
20. Phelps K, Taylor CKS, Nagel R, et al. Factors associated with emergency department utilization for non-urgent pediatric problems. Arch Fam Med 9 (2000): 1086-1092.
21. Nath JB, Costigan S, Lin F, et al. Federally Qualified Health Center access and emergency department use among children. Pediatrics 138 (2016): 20160479.