 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
,
,¶
* Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, Boston, Massachusetts
Department of Pediatrics, Harvard Medical School, Boston, Massachusetts
General Pediatrics
¶ Infectious Diseases, Department of Medicine, Children’s Hospital Boston, Boston, Massachusetts
|| Department of Pediatrics, University of Massachusetts School of Medicine, Worcester, Massachusetts
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Setting. Sixteen nonoverlapping Massachusetts communities enrolled in a community intervention study on appropriate antibiotic use.
Design. Pediatricians, family physicians, and a random sample of parents of children <6 years old were surveyed. Parents predicted what their satisfaction would be with initial observation of an ear infection without antibiotics if suggested by their physician and concerns they would have regarding this watchful-waiting approach. Physicians reported the frequency with which they use this approach in children 
2 years and those <2 years old. Separate multivariable models identified factors independently associated with parental satisfaction and with frequency of self-reported use by physicians. All models accounted for clustering of responses within communities.
Results. Two thousand fifty-four (40%) parents and 160 (58%) physicians responded. Of the parents, 34% would be somewhat or extremely satisfied if initial observation was recommended, another 26% would be neutral, and the remaining 40% would be somewhat or extremely dissatisfied. The multivariable model showed lower parental education (odds ratio [OR]: 0.50; 95% confidence interval [CI]: 0.35, 0.71, for high school education or less compared with college graduation) and Medicaid enrollment (OR: 0.77; CI: 0.57, 1.0) was associated with lower predicted satisfaction. Higher antibiotic-related knowledge (OR: 1.2; CI: 1.1, 1.3, per question correct), belief that antibiotic resistance is a serious problem (OR: 2.3; CI: 1.8, 2.8), and reporting feeling included in medical decisions (OR: 1.4; CI: 1.1, 1.7) all were independently associated with higher predicted satisfaction. Thirty-eight percent of physicians treating children 2 years old never or almost never reported using initial observation, 39% reported use occasionally, 17% sometimes, and 6% most of the time. In a multivariable model, only more years in practice (OR: 0.96; CI: 0.93, 0.99) was associated with a decreased likelihood of occasional or more-frequent use of watchful waiting (compared with those who never use initial observation). However, a secondary model that combined occasional users with nonusers (compared with those reporting use sometimes or more often) identified several correlates of use of observation: years in practice (OR: 0.95; CI: 0.91, 0.99), family medicine specialization (OR: 4.5; CI: 1.9, 11), belief that antibiotic resistance is a significant problem (OR: 4.3; CI: 1.3, 14.5), and practice in a community receiving a judicious antibiotic-use intervention (OR: 3.5; CI: 1.3, 9.1).
Conclusions. A majority of physicians reported at least occasionally using initial observation, but few use it frequently. Many parents have concerns regarding this option, but acceptability is increased among those with more education and those who feel included in medical decisions. Substantial change in both parental and provider views would be needed to make initial observation a widely used alternative for acute otitis media.
Key Words: otitis media • watchful waiting • observation option
Abbreviations: AAP, American Academy of Pediatrics • OR, odds ratio • CI, 95% confidence interval
The diagnosis of otitis media is the most frequent reason for prescribing antibiotics to young children1,2 and has, until recently, been rising rapidly.3,4 In March 2004, the American Academy of Pediatrics (AAP) released a practice guideline for the treatment of acute otitis media.5 It encourages careful diagnosis, use of narrow-spectrum antimicrobial agents, and allows, as an acceptable option, the initial observation of nonsevere cases of acute otitis media in selected children, including children 2 years old with mild ear pain, without high fever, and for whom the physician believes that prompt follow-up is assured should symptoms worsen.5 Support for this change has been catalyzed by increasing concern regarding antibiotic overuse and resistance6,7 and the realization that changes in diagnosis and/or prescribing patterns for otitis could substantially affect antibiotic-exposure rates in young children.8
Evidence for the safety and effectiveness of initial observation (also called "watchful waiting") comes from structured reviews of randomized drug trials that, despite their flaws,9 suggest that immediate antibiotic treatment has only a modest effect on the course of acute otitis media.10–12 A number of trials were conducted in the Netherlands,13,14 where watchful waiting is the standard treatment for most children with acute otitis media, but smaller trials of this option (with a "safety-net" prescription given to parents) have also been conducted in Britain15 and the United States.16 However, the observation option runs counter to the treatment paradigm for acute otitis media used for decades in the United States, and initial observation has been opposed by some US experts, who believe that early treatment is warranted to reduce symptoms and prevent serious complications such as mastoiditis.17
Initially observing cases of acute otitis media would represent a dramatic shift in treatment for an extremely common illness that is responsible for substantial morbidity and cost.18,19 It is not known how often primary care physicians in the United States are using this approach already or whether it will be acceptable to most parents. As part of a larger controlled trial, we surveyed both parents and providers in 16 Massachusetts communities, 8 of which had received a community-wide educational intervention to promote judicious antibiotic use. Through these simultaneous surveys, conducted before release of the AAP practice guideline, we sought to determine (1) the self-reported rates of current use of initial observation without antibiotics among physicians, (2) the acceptability to parents of this option, (3) the concerns of parents and physicians that influence choice of this strategy, and (4) physician factors associated with current use and parental factors associated with acceptability.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
2.0)21; these rates are consistent with national data on pneumococcal antibiotic resistance.22 Eight study communities had received an educational intervention for both patients and physicians over the prior 3 winter seasons to combat prevalent misconceptions about antibiotic use (REACH Mass [Reducing Antibiotics for Children]). The intervention included mailed newsletters, parent-education materials in physician waiting rooms, community-wide meetings with physicians, and feedback to physicians of current prescribing practices. Although not primarily focused on otitis media, the intervention provided basic information to parents regarding otitis media and its treatment in newsletters and waiting-room materials during the first 2 years. In the final year, the intervention included information about the "observation option" without specifically endorsing it.
Parent Survey
Parent surveys were mailed to a stratified random sample of 5580 commercially or Medicaid-insured parents (3194 commercial and 2386 Medicaid) who resided within a study community and had at least 1 child <6 years old. Addresses were obtained from 3 large commercial health plans and the state Medicaid program. An initial mailing and 2 follow-up mailings were sent between May 4, 2003, and July 30, 2003. The parent survey included demographic information including parental age, number of children, educational level, and race/ethnicity. It also included 10 previously reported general-knowledge questions about appropriate use of antibiotics23 and 3 items (such as: "If my doctor does not prescribe an antibiotic when I think one is needed, I will take my child to another doctor") previously used to identify parents with a proclivity to demand unnecessary antibiotics.24 The item used as the primary outcome for this analysis asked: "If your child’s doctor diagnosed an ear infection and recommended waiting 1 or 2 days before starting antibiotics (to see if the symptoms get better on their own) would you be...," with possible responses on a 5-point scale from extremely dissatisfied to extremely satisfied. Subsequent items queried parents about specific concerns regarding this strategy as well as their concerns about their child receiving a course of antibiotics.
Physician Survey
Surveys were sent in 2 mailings to 292 pediatricians and family physicians between April 30, 2003, and July 5, 2003. Items included respondents’ general level of concern about antibiotic resistance, perception of the prevalence of parent "demand" for unnecessary antibiotics, and self-report of recent changes in treatment of respiratory tract infections. Physicians were asked: "Among otherwise healthy children less than 2 years old...how often do you do the following for uncomplicated ear infections." They were asked to report the frequency with which they typically treat with standard-dose amoxicillin (45 mg/kg per day), high-dose amoxicillin (75–90 mg/kg per day), or "defer antibiotics for 24–72 hours and treat only if symptoms persist or worsen ("watchful waiting")." The response scale included "never/almost never," "occasionally," "sometimes," "most of the time," and "always/almost always." The question then was repeated, specifying children 2 years old with uncomplicated ear infections.
Analysis
We assessed comparability on key variables between intervention and control (nonintervention) communities by using the 
2 statistic or Fisher’s exact test for categorical variables and t test for continuous variables. Parent satisfaction, measured on a 5-point scale, was collapsed into a 3-level variable after examining the distribution of responses: satisfied, neutral, and dissatisfied. Bivariate statistics were used to identify parent factors associated with higher self-predicted satisfaction and to assess any impact of the intervention. Factors included both demographic attributes (not easily changed) and knowledge/belief variables (potentially modifiable through education or experience). Multivariable regression models were constructed to identify independent factors associated with a higher likelihood of satisfaction with the observation option among parents. A similar modeling approach was used to identify physician factors associated with current use of observation. The primary analysis identified factors that independently differentiated those who never use observation from those reporting use occasionally or more frequently. A secondary analysis combined occasional users with nonusers and identified factors distinguishing them from physicians who used observation sometimes, most of the time, or always. In all regression models, we accounted for nonindependence of observations within communities (clustering) (Glimmix macro, SAS 8.2, SAS Institute, Cary, NC). All variables with an association with the outcome at a level of P < .10 were entered into the multivariable models. Odds ratios (ORs) and 95% confidence intervals (CIs) are presented. This study was approved by the Human Subjects Committee at Harvard Pilgrim Health Care.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
Physician Survey
One hundred sixty physicians responded (response rate: 58%), of whom 117 (74%) were pediatricians and 60 (38%) were female. Respondents had practiced a mean of 17 years (SD: 10 years) (Table 4). Most respondents (60%) believed that antibiotic resistance was at least a moderately significant problem in their community. When presented with patients 2 years old, 100 (63%) reported at least occasionally using observation, including 27 (17%) reporting use sometimes and 10 (6%) most of the time. Among physicians who ever use watchful waiting among children 2 years old (n = 100), preferences of the family as well as clinical findings were commonly cited as influencing their decision (Table 5). Among physicians who never use initial observation (n = 60), the most commonly cited concerns were treatment failure (76%), speed of symptom resolution (68%), reduced patient satisfaction (61%), and risk of severe complications (56%). Less commonly cited were concerns that patients would not return for needed follow-up (27%), that observation was not (at that time) endorsed by professional societies (37%), or malpractice liability risk (36%).
|
|
2 years old (P < .01), with a marked difference in the fraction that use it most of the time (10% vs 0%). Physicians in intervention communities were less likely than physicians in control communities to be influenced by familiarity with the family (79% vs 94%; P = .04), family preferences (66% vs 94%; P < .01), severity of pain (73% vs 92%; P = .02), and magnitude of fever (72% vs 89%; P = .06). For children <2 years old, 63 (39%) reported at least occasionally using initial observation, including 15 (9%) who used observation sometimes, and 5 (3%) who used it most of the time (Table 5). There was a statistical trend toward more physicians in intervention communities using observation (P = .06) in these younger children.
The purpose of our primary analysis was to identify physician factors associated with at least occasionally using initial observation among children 2 years old compared with those who report never doing so. When analyzed this way, only more years in practice was independently associated with decreased likelihood of at least occasionally using observation (OR: 0.96; CI: 0.93, 0.99, for each additional year in practice). We conducted a secondary analysis to identify factors that might differentiate physicians who report using watchful waiting sometimes, most of the time, or all the time from those using it never or occasionally (Table 6). In this multivariable analysis, in addition to the negative correlation with years in practice (OR: 0.95; CI: 0.91, 0.99), family medicine specialization (OR: 4.5; CI: 1.9, 11.0), belief that antibiotic resistance is a very significant problem (OR: 4.3; CI: 1.3, 14.5), and practice in an intervention community (OR: 3.5; CI: 1.3, 9.1) were positively associated with use of the observation option.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Rising rates of antibiotic resistance in communities21 and concern about high rates of antibiotic use7 have led US physicians to consider adopting a strategy of initial observation in selected cases of acute otitis media, which is used successfully in other countries.14,25 Initial observation has appeal because of its potential to reduce antimicrobial use for an illness in which the benefits of immediate treatment are modest. Recent evidence-based reviews and meta-analyses suggest that 8 to 17 patients need to be treated to have a benefit (in symptoms measured at 2–7 days) in 1 patient.10–12 Clinical trials have demonstrated reasonable acceptability and outcomes with this strategy but, given their sample sizes, cannot rule out increased risk of rare but serious infectious outcomes such as mastoiditis.12 The potential advantages of initial observation are reflected in recent guidelines from the AAP, which endorses as an option initial observation of acute otitis media in patients 2 years old with nonsevere illness, in whom careful parental observation and follow-up are assured.5 In patients 6 months to 2 years old, observation is recommended as an option only in nonsevere cases in which the diagnosis is uncertain.
Our finding that most parents are not initially receptive to the observation option when presented as a vignette is not surprising. Most will have seen their children apparently respond to antibiotic treatment for acute otitis media, and only a small number will have ever had initial observation suggested by their child’s physician. Their level of comfort might be substantially higher if the option were explained by their own provider, or it might be lower in a real-life situation in which their child is ill. In either case, parental opinions in a community are likely to change as experience with successful treatment of acute otitis media without antibiotics becomes more common.
The data also suggest that the greatest challenge to the use of initial observation may be among less-educated or economically less-advantaged patients (for which insurance under Medicaid was a proxy in this study). Expectations for antibiotic treatment of upper respiratory illnesses have been shown to vary by patient race/ethnicity.26 It is possible that patients from underserved groups, for whom the quality of care is often lower,27 may be less accepting of an option that may be perceived initially to be "less" treatment. Alternatively, lower predicted acceptance may be driven by less knowledge (and therefore lower receptivity) or parental preference (even with good understanding of the pros and cons). Finally, our model suggests that parents who feel "included" in treatment decisions are more likely to be amenable to initial observation. Future work should consider whether economically disadvantaged parents or those of racial or ethnic minorities do, in fact, feel less included in decisions in the setting of acute childhood illnesses.
The 16 communities studied participated in a randomized trial to test a community-level intervention to promote judicious antibiotic prescribing for children. The intervention focused on the lack of benefit of antibiotics for colds, coughs, and flu-like illnesses and included education about otitis media. One newsletter provided information on treatment options for acute otitis media, including the observation approach, but did not advocate a particular strategy. This intervention did not seem to have an effect on parents’ receptivity to the observation option. Physicians in these communities were invited to evening educational sessions and received e-mailed "briefs" on topics related to antibiotic use and resistance. Data on observation of acute otitis media were presented, and patient-information handouts, based on a published toolkit,28 were distributed. The intervention did seem to have had a small effect on the number of physicians reporting use of the observation option in their practices.
The process of dissemination and adoption of the AAP guideline supporting initial observation provides a unique window on the diffusion of an innovation in medical practice. Speed of adoption by the majority is predicted to be faster when an innovation has great advantage for its users, is compatible with existing values and beliefs, has low complexity, and is easily testable by individuals.29,30 Even its advocates admit that the relative advantage of initial observation of ear infections is small to the individual physician and patient (although they posit longer-term community benefits of more judicious antibiotic use). Furthermore, some experts who serve as opinion leaders in the field do not believe that the evidence supports the effectiveness of watchful waiting at all.9,31 Finally, the recommendation to "observe" is, if anything, counter to the prevailing values of our medical care system, which tends toward therapeutic intervention. Although a group of "early adopters" exists already, these data show that they are strongly influenced by the perceived desires of their patients (even if physicians’ perceptions of patient desires are often wrong32). For all these reasons one would predict that the practice of initial observation will follow a slow adoption curve.
The opponents of the observation option criticize many of the studies on which current recommendations are based for including patients who likely had otitis media with effusion.9 They suggest that if stricter criteria were applied to diagnosing acute otitis media, as recommended by the AAP guidelines, antibiotic treatment would be reserved for infections more likely to be bacterial and would have greater effect. Data already exist suggesting that more severely affected children (with fever and vomiting) are more likely to benefit from early treatment.33 Additional work identifying reliable features that would identify patients who benefit most from antibiotic treatment is needed.34
Although the current data are from Massachusetts only, they represent views of patients and providers in 16 very different communities. Such data will serve as a baseline on which to judge the impact of new AAP/American Academy of Family Physicians guidelines supporting observation as an option in selected cases.5 As with any survey study, it is possible that the views of responders do not reflect those of the population overall. Especially because some associations were found in a secondary analysis that redefined the cutoff for physicians to be classified as "users," these findings require replication in other populations. Specifically, the finding that family physicians were more likely to use observation conflict with prior data from a Virginia study suggesting that pediatricians prescribed antibiotics more judiciously than family physicians.35 It would not be surprising if such differences varied depending on local and regional norms in training and practice. It is also possible that the relatively small number of family-physician respondents in our survey do not represent well the specialty as a whole. Finally, responses to vignettes in a survey (especially one in which a novel idea is presented) may underestimate the acceptability of the same approach were it presented by a trusted clinician with an opportunity for questions to be asked and answered.
Some have hailed release of the AAP guideline on acute otitis media as a way to further decrease unnecessary antibiotic use. Others have expressed concern about unintended consequences including increased frequency of serious bacterial infections. Ultimate adoption of initial observation will depend on additional evidence of its safety and efficacy in actual practice and clinical experience among community physicians. This study suggests that changes in practice will also depend considerably on parental knowledge and attitudes regarding the risks and benefits of antibiotics and the outcome of shared decision-making by patients and providers in the treatment of individual children.
|
ACKNOWLEDGMENTS |
|---|
We acknowledge the helpful comments provided by Richard Platt, MS, MD, and the support of the entire REACH Mass study team.
|
FOOTNOTES |
|---|
Address correspondence to Jonathan A. Finkelstein, MD, MPH, Department of Ambulatory Care and Prevention, Harvard Medical School, 133 Brookline Ave, 6th Floor, Boston, MA 02215. E-mail: jonathan_finkelstein{at}harvardpilgrim.org
No conflict of interest declared.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. A. Finkelstein, S. S. Huang, K. Kleinman, S. L. Rifas-Shiman, C. J. Stille, J. Daniel, N. Schiff, R. Steingard, S. B. Soumerai, D. Ross-Degnan, et al. Impact of a 16-Community Trial to Promote Judicious Antibiotic Use in Massachusetts Pediatrics, January 1, 2008; 121(1): e15 - e23. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Vernacchio, R. M. Vezina, and A. A. Mitchell Management of Acute Otitis Media by Primary Care Physicians: Trends Since the Release of the 2004 American Academy of Pediatrics/American Academy of Family Physicians Clinical Practice Guideline Pediatrics, August 1, 2007; 120(2): 281 - 287. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. M. Spiro, K.-Y. Tay, D. H. Arnold, J. D. Dziura, M. D. Baker, and E. D. Shapiro Wait-and-see prescription for the treatment of acute otitis media: a randomized controlled trial. JAMA, September 13, 2006; 296(10): 1235 - 1241. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. A. Varrasso, D. Ashe, R. Ruben, and R. Propp Watchful Waiting for Acute Otitis Media: Are Parents and Physicians Ready? Pediatrics, August 1, 2006; 118(2): 849 - 850. [Full Text] [PDF]
|
|
|
|
 |
|
 Minerva BMJ, September 17, 2005; 331(7517): E382 - E382. [Full Text] [PDF]
|
|
|
|
|
|
Minerva BMJ, June 25, 2005; 330(7506): 1516 - 1516. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2005 American Academy of Pediatrics. All rights reserved. |
||
|
||
