An evidence based approach to reducing antibiotic use in children with acute otitis media: controlled before and after study

Two recent reviews questioned the routine use of antibiotics in the initial management of acute otitis media.¹,² My practice partners and I responded to the reviews by changing our policy, with the aim of reducing such prescribing in children. We measured the change one year after adopting the new policy.

Methods and results

From July 1997 my practice partners and I changed our policy on routine prescription of antibiotics in the initial management of acute otitis media in children. In children who were not particularly ill we gave the parents a handout that summarised the limited benefit of antibiotics on the basis of the data presented in the Cochrane review.¹ We advised parents to give regular paracetamol suspension; we also offered an antibiotic prescription but asked the parents to keep it for a day or two. They could redeem it at a pharmacy if the child did not got better over this period.

A local practice acted as a concurrent control. Both practices use amoxicillin suspension as the antibiotic of choice in children with acute otitis media, and, although the doctors in the control practice were aware of the new evidence, they did not use the handout or use deferred prescriptions. Monthly prescribing rates of all amoxicillin suspensions were obtained for each practice from the district health authority. The 12 months before July 1997 were used for baseline comparison, and the following 12 months were used to assess the impact of our change in policy. Both practices had closely similar list sizes (about 11 000 patients) throughout the study. As there was seasonal and annual variation in prescribing levels, monthly odds of prescriptions issued in relation to the national total were calculated for each practice; these were weighted and pooled by using the Mantel-Haenszel method.³

Prescriptions for other antibiotic suspensions were also checked in our practice by comparing figures from the reports on level 3 prescribing analysis and cost data (PACT) with national figures, to check that other antibiotics were not being substituted for amoxicillin.

The table shows the monthly prescriptions for amoxycillin suspension for each practice, along with national totals. The median number of prescriptions per month in our practice fell from 75 to 47 after the change (median difference −30.5 (95% confidence interval −14 to −31, 2P=0.0065, Mann-Whitney U test). Compared with the national levels, the fall in prescribing amoxicillin suspension in our practice was −32% (−25% to −39%) and in the control practice was −12% (−4% to −20%).

In the six months after the change in policy the number of antibiotic suspensions prescribed in our practice was 19% lower (16% to 21%) than in the same six months of the previous year. Over this period national prescribing of all antibiotic suspensions fell by 3%.⁴

Comment

Before my partners and I changed our policy, acute otitis media accounted for over half of all antibiotics prescribed for children in our practice. After our change in policy the proportion fell to a third, and as a consequence total antibiotic use for all infections in childhood fell by one fifth. As the evidence is not yet available to identify which children with acute otitis media benefit most from antibiotics we deferred prescriptions in those who were not systemically unwell (as suggested in the editorial comment accompanying one of the reviews in the BMJ ⁵).

We found that most parents welcomed the written handout. The deferred prescription also acted as a safety net while they waited to see if the ear infection would resolve by itself and was often not redeemed. We are continuing with this policy, and in the current battle against antimicrobial resistance we would commend this approach to initial management of acute otitis media as a way of reducing the antibiotic load on children in the community.

Table.

Monthly totals of prescriptions for all amoxicillin suspensions for 12 months before (1996-7) and after (1997-8) new prescribing policy was introduced

Month	Practice using new policy			Control practice			Nationally
	Before	After		Before	After		Before	After
July	68	55		68	66		392 367	364 831
August	35	31		35	26		220 606	214 292
September	74	33		56	40		367 901	366 957
October	65	45		72	54		429 828	503 459
November	106	62		150	90		502 174	530 556
December	162	117		169	155		869 621	752 960
January	95	49		85	72		634 195	489 216
February	86	78		83	91		538 509	591 498
March	80	49		86	74		470 010	515528
April	70	44		64	66		392 452	354 032
May	76	39		72	50		442 111	323 566
June	71	37		51	51		345 765	369 194
Median	75	47		72	66		435 970	429 205
% change from previous year (95% CI)*	−32% (−39% to −25%)			−12% (−20% to −4%)

^*Change in pooled odds ratios for the odds of prescriptions issued each month related to the national total for that month and compared with the odds for the same month in the previous year (the Mantel-Haenszel method was used to weight and pool the odds ratios).³