Otitis media, tympanostomy tube placement, and use of antibiotics

Abstract

Objective. To investigate potential links between antimicrobial drug use for acute otitis media (AOM) and tympanostomy tube placements, and the relationship between parental views and physician antimicrobial prescribing habits. Design. Cross-sectional community study repeated after five years. Subjects. Representative samples of children aged 1–6 years in four well-defined communities in Iceland, examined in 2003 (n = 889) and 1998 (n = 804). Main outcome measures. Prevalence of antimicrobial treatments for AOM, tympanostomy tube placements, and parental expectations of antimicrobial treatment. Results. Tympanostomy tubes had been placed at some time in 34% of children in 2003, as compared with 30% in 1998. A statistically significant association was found between tympanostomy tube placement rate and antimicrobial use for AOM in 2003. In the area where antimicrobial use for AOM was lowest in 1998, drug use had further diminished significantly. At the same time, the prevalence of tympanostomy tube placements had diminished from 26% to 17%. Tube placements had increased significantly, from 35% to 44%, in the area where antimicrobial use for AOM was highest. Parents in the area where antimicrobial consumption was lowest and narrow spectrum antimicrobials were most often used were less likely to be in favour of antimicrobial treatment. Conclusions. Comparison between communities showed a positive correlation between antimicrobial use for AOM and tympanostomy tube placements. The study supports a restrictive policy in relation to prescriptions of antibiotics for AOM. It also indicates that well-informed parents predict a restrictive prescription policy.

• Acute otitis media
• antimicrobial use
• drug prescription
• epidemiology
• Iceland
• intervention
• parental expectations
• tympanostomy tube placement

Acute otitis media (AOM) is the single most common diagnosis resulting in antimicrobial prescription for children in the Western world [1–4], including Iceland. Nevertheless antimicrobial treatment of AOM is debatable. Several randomized clinical trials [5], [6] and meta-analyses [7], [8] indicate that it may often be unnecessary, yet severe infections such as acute mastoiditis can complicate AOM. Comparison of the incidence of mastoiditis in the USA where AOM has usually been treated with antimicrobials and the Netherlands where AOM is usually not treated with antimicrobials, however, reveals a negligible difference [9]. A restrictive antibiotic policy for AOM therefore appears justifiable. Such a policy is further supported by the documented link between high consumption of antimicrobials and development of antimicrobial resistance among common pathogens [1], [10], [11]. A strong association between antimicrobial use for AOM and carriage of penicillin non-susceptible pneumococci in children has been demonstrated [12]. Further knowledge about the short- and long-term impact of different treatment strategies for otitis media is important.

Tympanostomy tube placement is a well established treatment option for chronic effusion in the middle ear and recurrent episodes of AOM [13], yet the consequences of this intervention are not clear [14–16]. The cumulative prevalence of tympanostomy tube placement among Icelandic pre-school children has been shown to be 30% [2], which is much higher than reported in other countries [17]. Furthermore, children with tympanostomy tubes received as many antimicrobial courses as children without tubes.

The aim of the present study was to investigate potential links between antimicrobial treatment for AOM and tympanostomy tube placements and evaluate the impact of sociodemographic factors and parental views on physician prescribing habits. We approached these research questions by conducting a cross-sectional investigation in four well-defined areas with a five-year interval.

Material and methods

Study population and design

Eligible participants included all children 1 to 6 years of age living in the four Icelandic communities Hafnarfjördur (H), Vestmannaeyjar (V), Egilsstadir (E), and Bolungarvik (B). All eligible children in V, E, and B were invited to participate, whilst a random sample was invited in area H (Table I). The study areas were served by a total of 19 general practitioners (GPs) (18.2 positions), of whom 11 were the same individuals as in the 1998 study. The organizational structure of the healthcare service was similar for both study periods, except in area H which is a suburb of Reykjavik. Emergency appointments were more often provided by doctors outside area H in 2003 than in 1998, resulting in somewhat declining continuity of care in area H. The same ear, nose, and throat specialists were responsible for bi-monthly visits to each of the areas V, E, and B during the years 1998 and 2003. Inhabitants in area H had easy access to such specialists in Reykjavik. Informed parental consent was obtained for all participants. The parents received a questionnaire regarding sociodemographic factors, number of children and smoking habits in the household, and day care attendance. Antimicrobial use was measured as the number of oral antimicrobial courses during the preceding year, according to parents’ recollection. The date of the last antimicrobial prescription was noted, together with the type of agent prescribed and the clinical indication (diagnosis) for the prescription. A history of past or present use of tympanostomy tubes was recorded. Information gathered from parents was matched with information from the medical records at the local health centre. Parents were furthermore asked to complete a questionnaire about their personal views in relation to antimicrobial treatment, similar to the 1998 study [2]. The study was approved by the National Bioethics Committee and the National Data Protection Commission, Iceland.

Table I.  Study population: characteristics of the populations in 2003 and 1998.

Area	Year	Total no. of inhabitants and no. of GP positions within parentheses	Children 1–6 years old No.	Children 1–6 years old invited No.	Participants No.	Participation rate	Mean age of participants (years±SD)
H Hafnarfjordur	1998	18.203 (9.2)	1.989	442	331	75%	4.07±1.75
	2003	20.675 (10.2)	2.004	380	287	76%	4.16±1.65
V Vestmannaeyjar	1998	4.640 (4)	512	321	244	76%	4.20±1.67
	2003	4.416 (4)	406	396	324	82%	4.37±1.57
E Egilsstadir	1998	3.076 (3)	300	198	174	88%	4.04±1.78
	2003	2.874 (3)	265	256	216	84%	4.20±1.71
B Bolungarvik	1998	1.094 (1)	99	66	55	83%	4.23±1.72
	2003	957 (1)	82	75	62	83%	3.89±1.64
Total	1998	27.013 (17.2)	2.900	1.030	804	78%	4.11±1.73
	2003	28.922 (18.2)	2.757	1.107	889	80%	4.22±1.64

Statistical analysis

The statistical methods have been described in detail elsewhere [2]. Fisher's exact test and the Pearson chi-square test were used for univariate analysis of categorical variables. Linear regression analysis was used to estimate correlation coefficient (r) and 95% confidence interval for the slope of mean number of antimicrobial courses for AOM compared with tympanostomy tube placement rate by areas and mean number of total antimicrobial courses compared with the prevalence of parents finding it appropriate sometimes to treat common cold with antimicrobials. Multivariate logistic regression models were used to estimate odds ratio. Individual regression coefficients were tested with the Wald statistic and 95% confidence intervals were calculated for the odds. The statistical analyses were performed with the SPSS® program for Windows (Advanced Statistics Release 8.0, SPSS Inc, Chicago, Ill.). The level of statistical significance was taken as 0.05.

Results

Participation

In total, 889 children participated (see Table I), representing a response rate from 76 to 84 percent. According to the questionnaires most of the children attended day care centres (84–94%). The child's mother provided the parental information in 90% of cases.

Prescription of antibiotics

A total of 48% of the participating children had received oral antimicrobial treatment at least once during the preceding 12 months, compared with 55% in 1998 (p = 0.007). The mean number of antibiotic treatments per year was 1.06 (1.21 in 1998, Table II). The usage was significantly higher in area V compared with the other areas on both occasions. The proportion of antimicrobial prescriptions for AOM was about the same (57%) as in 1998 (58%). The number of treatment courses for the indication AOM was also highest in area V, i.e. 0.88 courses per year per child (Table II). In area E, where the antimicrobial prescription rate was lowest (0.57) in 1998, it had decreased significantly to 0.27 in 2003.

Table II.  Total number of antimicrobial treatment courses during the 12 months prior to enrolment, number of treatment courses for AOM, and number (%) of children 1–6 years of age having had tympanostomy tube placement.

Area	Year	No. of treatment courses per child per year (95% CI)	No. of treatment courses for AOM per child per year (95%CI)	No. (%) of children having had tympanostomy tube placement	p-value* for changes in tympanostomy tube placement rate between 1998 and 2003
H	1998	1.16 (0.99–1.33)	0.69 (0.56–0.83)	81/310 (26.1)
	2003	1.05 (0.86–1.23)	0.57 (0.43–0.71)	103/280 (36.8)	0.006
V	1998	1.51 (1.26–1.75)	0.81 (0.62–1.00)	79/227 (34.8)
	2003	1.48 (1.27–1.69)	0.88 (0.71–1.04)	143/324 (44.1)	0.034
E	1998	0.89 (0.60–1.09)	0.57 (0.41–0.74)	39/148 (26.4)
	2003	0.53 (0.41–0.65)	0.27 (0.19–0.36)	37/215 (17.2)	0.049
B	1998	1.15 (0.72–1.57)	0.65 (0.32–0.99)	17/50 (34.0)
	2003	0.72 (0.46–0.98)	0.41 (0.20–0.62)	15/61 (24.6)	0.299
Total	1998	1.21 (1.09–1.32)	0.70 (0.61–0.79)	216/735 (29.4)
	2003	1.06 (0.95–1.16)	0.60 (0.52–0.68)	298/880 (33.9)	0.060

Corresponding figures from the 1998 study are included.

The choice of antimicrobial drug also varied considerably by study areas. Broad-spectrum antimicrobials (amoxicillin + clavulanic acid, azitromyzin, erythromycin, trimethoprim-sulphamethoxazole), were used in only 23 of 72 cases (32%) in area E where the total antimicrobial consumption was lowest but in 148 of 189 cases (78%) in the high-consumption area V (p < 0.0001).

Predictors of prescriptions for antibiotics for otitis media

As shown in Table III, the child's age, residence, and history of tympanostomy tubes (past or present) were independently associated with the number of antimicrobial treatments for AOM. In our 1998 study, this was the case only for age. Smoking did not influence the odds if the child received antimicrobials for AOM or not.

Table III.  Mean number of antimicrobial treatment courses per year for AOM and adjusted odds ratios (OR) for children treated with antimicrobials for AOM in the preceding 12 months (n = 270), as opposed to not (n = 593).

Variable	Mean number of treatment courses per year for AOM (95%CI)	OR	95% CI	p-value
Age
1–3 yrs	1.06 (0.90–1.21)	1
4–6 yrs	0.23 (0.18–0.28)	0.18	0.12–0.25	0.0001
Gender
male	0.61 (0.50–0.72)	1
female	0.59 (0.47–0.70)	1.01	0.73–1.40	0.99
Siblings
No	0.82 (0.57–1.08)	1
Yes	0.55 (0.47–0.63)	0.79	0.52–1.20	0.25
Residence
Area V	0.88 (0.71–1.04)	1
Area E	0.27 (0.19–0.36)	0.32	0.20–0.51	0.0001
Area B	0.41 (0.20–0.62)	0.44	0.22–0.88	0.019
Area H	0.57 (0.43–0.71)	0.50	0.34–0.74	0.0003
Day care
No	0.89 (0.58–1.20)	1
Yes	0.57 (0.49–0.66)	0.86	0.50–1.47	0.56
Tympanostomy tube, now or before
Yes	0.94 (0.75–1.13)	1
No	0.43 (0.36–0.50)	0.49	0.35–0.70	0.0001
Smoking in the house
No	0.59 (0.50–0.68)	1
Yes	0.64 (0.45–0.82)	0.92	0.61–1.41	0.69

Multivariate analysis (863 cases analysed, data missing for 26 cases). Overall significance: chi-square = 165,79, df = 10, p = 0.0000.

Predictors of tympanostomy tube placements

Information about tympanostomy tube usage was obtained for 880 children. A total of 298 (34%) reported tympanostomy tube insertions in one or both ears at some time point (cumulative prevalence), and among these, as many as 22% (95% CI 14–31%) of one-year-old children. Boys, as in 1998, had received tympanostomy tubes significantly more often (178/467) than girls (120/413) (p = 0.018). The percentage of children who had received tubes varied significantly from 17% in area E to 44% in area V in 2003 (see Table II, p < 0.0001).

From 1998 to 2003, there had been a significant increase in tympanostomy tube placements in areas V and H (where antimicrobial prescription rates were highest), and a decrease in areas E and B (not significant in area B) (see Table II). In area E, a concurrent and significant decrease took place in tympanostomy tube placements among children and antimicrobial prescriptions (total use as well as for the indication AOM specifically). According to multivariate analysis, adjusted for age, a child having tympanostomy tube(s) was more likely to have had recurrent antimicrobial treatment and to live in the area where antimicrobials were most frequently used (area V, Table IV). Day care attendance was also a predictor of having tubes. A significant correlation was found at the community level between tympanostomy tube placement rate and antimicrobial use for AOM (r = 0.97, p = 0.032) (Figure 1).

Figure 1. Correlation between tympanostomy tube placement rate (cumulative prevalence) and antimicrobial use for acute otitis media (last 12 months) among 1- to 6-year-old children in Iceland, 2003.

Table IV.  Odds ratio (OR) adjusted for age for children having tympanostomy tubes (n = 123), as opposed to not (n = 736).

Variable	No.	OR	95%CI	p-value
Residence
Area V	314	1
Area E	214	0.35	0.17–0.64	0.0021
Area B	61	1.02	0.46–2.24	0.97
Area H	270	0.81	0.50–1.30	0.38
Number of antimicrobial treatments for AOM, last 12 months
0	594	1
1–2	203	1.96	1.22–3.17	0.0059
≥ 3	62	4.98	2.60–9.53	0.0001
Gender
Female	407	1
Male	452	1.28	0.84–1.94	0.25
Smoking in the family house setting
No	701	1
Yes	158	1.01	0.60–1.72	0.95
Day care
No	71	1
Yes	788	2.38	1.07–5.28	0.0317

Multivariate analysis (859 cases analysed, data missing in 30 cases). Overall significance: chi-square = 100.176, df = 9, p = 0.0000.

Effects of tympanostomy tubes on recent antimicrobial drug consumption

Of 123 children having tubes, 49 (40%) had received oral antimicrobials at least once during the previous 8 weeks, as compared with 124 of 737 children (17%) without tubes (p < 0.001). Considering only antimicrobial treatment for AMO, 25 children (20%) in the tympanostomy group had recently received antimicrobials, compared with 74 children (10%) among those without tubes (p < 0.001). Adjusting for age, the odds ratio for receiving antimicrobials was significantly higher for children with tubes (p < 0.001). In 1998, this difference was not statistically significant.

Parental views on antimicrobial drug treatment

Table V gives the results of the questionnaires regarding appropriate antimicrobial usage for upper respiratory infections in children. Data from 1998 are provided for comparison. At the community level, a significant correlation was found between the children's total antimicrobial use and the rate of parents considering it sometimes appropriate to treat common cold with antimicrobials (r = 0.99, p = 0.01).

Table V.  Results from parents’ questionnaires regarding appropriate antimicrobial usage for upper respiratory tract infections for children, number and percentage (%), compared with results from the 1998 study.

Question/area	Year	Always	Sometimes	Never	Don't know
Question 1^1		Is it appropiate to give antimicrobial treatment for acute otitis media?
Area H	1998	27 (8.3)	286 (87.5)	3 (0.9)	11 (3.4)
	2003	30 (10.7)	231 (82.5)	3 (1.1)	16 (5.7)
Area V	1998	28 (11.5)	197 (81.1)	4 (1.6)	14 (5.8)
	2003	33 (10.2)	274 (85.1)	1 (0.3)	14 (4.3)
Area E	1998	9 (5.2)	152 (87.9)	5 (2.9)	7 (4.0)
	2003	11 (5.1)	194 (90.7)	1 (0.5)	8 (3.7)
Area B	1998	2 (3.6)	51 (92.7)	1 (1.8)	1 (1.8)
	2003	1 (1.6)	58 (95.1)	1 (1.6)	1 (1.6)
Total	1998	66 (8.3)	686 (86.0)	13 (1.6)	33 (4.1)
	2003	75 (8.6)	757 (86.3)	6 (0.7)	39 (4.4)
Question 2^2		Is it appropriate to give antimicrobial treatment for unspecified upper respiratory tract infection (common cold)?
Area H	1998	0 (0.0)	154 (47.2)	148 (45.4)	24 (7.4)
	2003	0 (0)	122 (44.0)	137 (49.5)	18 (6.5)
Area V	1998	1 (0.4)	121 (49.8)	95 (39.1)	26 (10.7)
	2003	0 (0)	186 (57.9)	114 (35.5)	21 (6.5)
Area E	1998	0 (0.0)	50 (28.9)	109 (63.0)	14 (8.1)
	2003	0 (0)	64 (30.2)	131 (61.8)	17 (8.0)
Area B	1998	0 (0.0)	21 (38.2)	25 (45.5)	9 (16.4)
	2003	0 (0)	20 (32.8)	38 (62.3)	3 (5.1)
Total	1998	1 (0.1)	346 (43.4)	377 (47.3)	73 (9.2)
	2003	0 (0)	392 (45.0)	420 (48.2)	59 (6.8)

¹1998, Pearson chi-square 13.01, df 9, p = 0.159; 2003, Pearson chi-square 15.05, df 9, p = 0.090. ²1998, Pearson chi-square 32.27, df 9, p = 0.0001; 2003, Pearson chi-square 46.65, df 9, p = 0.0001.

Discussion

Four communities in Iceland have been studied with an interval of five years with regard to use of antimicrobial drugs, prevalence of tympanostomy tube placements, and prevalence of penicillin non-susceptible pneumococci, in particular to see how these factors interact with each other. Our earlier studies have highlighted the risk of recent antimicrobial usage on carriage of penicillin non-susceptible pneumococci [1], [12] as well as an unusually high prevalence of tympanostomy tube placements in Iceland [2]. In the present paper we observed not only an association between antimicrobial prescription rates for AOM and prevalence of tympanostomy tube placements at the community level, but also changes with time between 1998 and 2003 in relation to antimicrobial usage. In the area where antimicrobial consumption was lowest in 1998, antimicrobial use for AOM as well as tympanostomy tube placements decreased significantly. In the two areas where antimicrobial use was highest in 1998, antimicrobial use remained high, and the tympanostomy tube placement rate increased significantly. Changes in the delivery of acute healthcare services in area H from 1998 to 2003 weakened the quality of the local data where a higher antimicrobial prescription rate could be expected.

There was a significant association between the number of antimicrobial prescriptions for AOM in children and tympanostomy tube placements on the level of both the individual and the community. This phenomenon may have different underlying explanations. One possibility is that frequent use of antimicrobial drugs might change the bacterial flora in the community, by promoting the emergence of resistant bacteria [1], [12]. This could in turn increase the risk of treatment-resistant episodes of bacterial infections including AMO [18], chronic effusion in the middle ear, and subsequent tympanostomy tube placement. A second possibility may be due to the fact that otorrhea is a common problem in children with tympanostomy tubes [19]. This may lead to prescriptions of antibiotics due to suspicion or fear of AOM and may at least partly explain why children with tubes in the present study had more often received recent oral antimicrobial treatment.

On the other hand, the observed difference of practice may have a complex set of explanations. The incidence of respiratory tract infections may vary between areas. There may also be a tendency to under-diagnose AOM in some areas and over-diagnose AOM in others. Furthermore, the various areas may have different treatment traditions/cultures in relation to the clinical problem in question. A recent study by Gabbay and Le May addressed the phenomenon of varying clinical “mindlines” among primary health care practices in the UK [20]. Parents in areas where antimicrobial drug usage was lowest and broad-spectrum antimicrobials most rarely prescribed appeared to be better informed about the benefits of restrictive use of antimicrobial drugs for upper respiratory tract infections. These variations in parental attitude may be important as primary care physicians maintain that parental pressure and expectations in relation to antimicrobial drug prescriptions represent an important barrier to a restrictive prescribing practice [2], [21–27].

Our results demonstrate an association between the use of antibiotics, parental views about antimicrobial treatment, and use of tympanostomy tube placements. However, it is impossible to say whether the associations between antimicrobial use for AOM and tympanostomy tube placements are due to direct causal effects or because of differences in medical practice. The possibility of a direct positive correlation between antimicrobial drug (over)use for AOM, future episodes of AOM, and tympanostomy tube placement later warrants further investigation.

In Iceland, 30% of pre-school children have had tympanostomy tube placements for recurrent or chronic otitis media. It is unclear whether this is due to high disease prevalence or medical over treatment.

• We found a positive correlation between tympanostomy tube placements and rate of antimicrobial drug prescription for acute otitis media.

• In regions with diminishing use of antimicrobials, a decrease in tympanostomy placements could be observed over time.

• Parental knowledge about appropriate use of antimicrobial drugs was found to be related to physician prescribing habits.

The authors would like to thank all the children and parents who participated in the study and the general practitioners who assisted in the study areas. This study was supported by the Icelandic College of Family Physicians and the Research Fund of the University of Iceland, Reykjavik.