Duration of intravenous antibiotic therapy for children with acute osteomyelitis or septic arthritis: a feasibility study

Osteomyelitis

Surgical drainage in acute OM is indicated if the patient is not responding to antibiotics after 48–72 hours (although this may be because of resistance) or if there is radiological evidence of a substantial pus collection. 7 Best practice is to immobilise any surgically treated limb or focus of infection. Occasionally, where a soft tissue or subperiosteal collection is clearly demonstrated by ultrasonography or MRI, needle aspiration can be performed prior to starting intravenous (i.v.) antibiotics. When performed, the procedure should be carried out under sterile conditions. If there is bony destruction or pus aspirated, surgical debridement is usually required. With only early radiographic signs, conservative i.v. antibiotic therapy may suffice.

Historically, the role of surgery is poorly defined. Cole et al. 42 identified three groups of patients. In the group of patients aged > 1 year who presented within 48 hours, antibiotic therapy alone was sufficient. In the group aged > 1 year, 5 days after the onset of illness, patients usually required surgery and possibly multiple procedures. In infants aged < 1 year, in whom the exact diagnosis was difficult to make, a single operation and antibiotic therapy usually sufficed. In current practice, the relative roles of bacterial virulence and host age and immunity are unclear. More invasive surgery appears more common when bacteria have specific virulence genes, for example PVL. 22 Although most children recover rapidly with simple medical management, others require repeated debridement.

Septic arthritis

In SA, prompt drainage and washout of the affected joint (either arthroscopic or open) is advocated by some for both diagnostic and therapeutic purposes as the articular cartilage is damaged early. 7 The role of surgery in the treatment of SA is, in fact, poorly defined except in relation to the hip, where prompt surgical drainage is absolutely necessary. Open capsulotomy to allow continuing drainage of septic material is advocated, and, if the arthrotomy does not provide turbid material, drilling the femoral neck may decompress a proximal femoral OM. The anterior approach is preferred as this also allows open reduction of any displacement of the femoral head.

The indications for surgical drainage of septic joints other than the hip remain controversial. Where there is a large effusion, drainage is usually advocated, although in some joints arthroscopic irrigation may be appropriate, such as the knee or ankle. However, with arthroscopic treatment joint visualisation is less complete. Overall, for joints other than the hip, aspiration, irrigation and i.v. antibiotic therapy is the preferred first line of treatment. If the patient fails to respond, then the joint should be surgically drained, usually by formal open arthrotomy rather than arthroscopic drainage.

Current evidence for how to initiate treatment

Intravenous antibiotics are started empirically as soon as the clinical diagnosis of acute OM or SA is made, as delaying therapy until the bacterium is identified increases the risk of complications. In SA, where urgent surgery is indicated, a widespread pragmatic approach has been to start antibiotics following surgery unless it will take > 4 hours to get to theatre. As soon as organisms are isolated, antimicrobial treatment should be adjusted and optimised. In subacute OM with no systemic reaction, oral antibiotics can be used from the start.

Although there has not been a definitive randomised controlled trial (RCT), a number of observational and retrospective studies in the literature show that several different antibiotic regimes have been effective in treating acute haematogenous OM in children, including the use of beta-lactam and macrolide antibiotics. 9

The initial antibiotics should always include potent cover against MSSA and GAS, and in younger children against K. kingae, although the choice will vary according to the age of the child, route of infection and local resistance patterns. 8 Recent data from the USA suggest increases in resistant S. aureus (MRSA), but these data have not been replicated in UK cohorts. 43 Activity against H. influenzae type b is essential in children who have not been fully immunised against it.

Switch to oral antibiotics and total duration of treatment

Currently there is no international and little UK consensus regarding the route or duration for antibiotic treatment of acute OAI in children.

Simple cases

Children presenting with SA or OM for the first time, and with no chronic comorbidities. Specifically:

• no orthopaedic surgery 6 months prior to presentation on same limb/joint

• no chronic comorbidities (sickle cell disease, known immunocompromise, indwelling central venous catheter or PICC, cystic fibrosis or cerebral palsy).

Complex cases

Children presenting with SA or OM who:

• have had any orthopaedic surgery in the 6 months prior to presentation on same limb/joint

• have chronic comorbidities (sickle cell disease, known immunocompromise, indwelling central venous catheter or PICC, cystic fibrosis or cerebral palsy)

• are presenting because of an infected implant from previous orthopaedic surgery.

Misdiagnosed

Some children captured by the database were treated for SA/OM but were later found to have a different diagnosis.

Other

Some children had discitis. This is a form of SA, but it is treated differently and so is not of interest to this study.

Participating centres

Fifty-one centres (49 hospital trusts) took part in the study. Eighteen of these were classed as tertiary centres and 33 as secondary centres.

Together, centres provided data for 356 children. In addition to these, three children were withdrawn as consent was withdrawn and five were excluded from analysis on the basis of a major lack of data. One tertiary centre (Chelsea and Westminster) and six secondary centres did not record any cases. For one of the secondary centres this was because the centre withdrew from the study.

Tertiary centres contributed 68% of the 356 cases (66% of all 218 simple cases). There was wide variation in the proportion of cases presenting with symptoms of OM/SA that were complex. In tertiary centres this varied from 0% to 75% (Table 1) and in secondary centres from 0% to 66.7% (Table 2).

Recruitment

The study opened to recruitment in June 2013 and closed 1 year later. During this time frame, sites recorded all children presenting with SA/OM. Sites did not all open for recruitment at the same time but most achieved a minimum of 6 months (with the exception of one tertiary centre and six secondary centres). The median [interquartile range (IQR)] recruitment time was 8.8 (6–9.4) months for tertiary centres and 8.5 (6.3–9.3) months for secondary care centres (see Tables 1 and 2).

Accrual rates for simple and complex cases

The monthly rates of case accrual are given in Tables 1 and 2. The median recruitment rate in tertiary centres was 0.7 (IQR 0.2–1.7) cases per month for simple cases and 0.4 (IQR 0.2–0.7) cases per month for complex cases. For secondary centres the median recruitment rate was 0.2 (IQR 0.1–0.5) simple cases per month and 0 (IQR 0–0.1) complex cases per month.

Initial categorisation of types of cases and sample characteristics

A total of 356 children were recorded in the database as presenting with symptoms of SA or OM. Two hundred and eighteen (61.2%) children were classified as simple cases of OM or SA, 95 (26.7%) were classified as complex cases, five (1.4%) had discitis and 38 (10.7%) presented with the symptoms of OM or SA but were found to have a different diagnosis. Discitis is a special case of SA, but it is treated in a different way and so is not of interest to this study.

Table 3 gives the breakdown of how many children had OM or SA in each case-type group and the types of diagnosis of misdiagnosed children. Note that, for six children, either there was evidence of both SA and OM, or the diagnosis was ambiguous in the database, meaning that it could have been either SA or OM, or both. These children are not included in the analyses split by diagnosis group.

A total of 53.9% of participants were male, the median age was 3.4 years and most were of north European, east European or mid-European origin (73.3%). Table 4 shows the sex, age group and ethnicity for all 356 children, and also for the 313 children who were diagnosed with SA or OM. When misdiagnosed children and children with discitis are excluded, the demography of the sample remains largely unchanged.

Completeness of follow-up

Children were followed up for approximately 3 months post discharge. A total of 72% of simple cases and 80% of complex cases were contactable and followed up. In most cases, failure to follow up was because children were not contactable (Table 5). In 2.8% and 2.1% of simple and complex cases, respectively, follow-up data were missing for unknown reasons.

Medical history

Presentation

Blood tests at presentation

Clinical timeline

All children in the study spent time in hospital. On average, children went straight to hospital (median time from presentation to hospital referral was 0 days), and most children presented at accident and emergency (A&E) (see Table 12).

Not all children received i.v. antibiotics (see Table 31). Fifteen (6.9%) simple cases and two (2.1%) complex cases had either no i.v. treatment or missing data. Of those who did receive i.v. antibiotics, we can use the date of initiation of i.v. therapy as a proxy for date of diagnosis.

The clinical timeline for first diagnostic X-ray, radiological diagnosis, first diagnostic surgery and the initiation of i.v. antibiotics are given in Table 17. A total of 61% of simple cases and 59% of complex cases had at least one X-ray post presentation and prior to initiation of i.v. antibiotics (i.e. had a diagnostic X-ray). These were usually on the day of or day after presentation (see Table 17).

Diagnostic surgery was carried out in 28% of simple cases at a median of 1 day post presentation (25% carried out at ≥ 4 days post presentation). For complex cases, the surgery was slightly more likely (30.5%), at a median of 1 day post presentation (25% carried out ≥ 5 days post presentation).

A radiological diagnosis [an abnormal result via any imaging method (X-ray, ultrasonography, MRI, bone scan or CT) post presentation and prior to initiation of i.v. antibiotics] was obtained for 51.8% of simple cases (at a median of 1 day post presentation) and 52.6% of complex cases (at a median of 1 day post presentation).

Diagnosis

Day of diagnosis was defined as the day when i.v. antibiotics were initiated as the standard of care has always been for i.v. antibiotics to be started as soon as possible in paediatric OM/SA. This can be pinpointed for 203 (93.1%) simple cases and 91 (94.6%) complex cases (see Table 31 for more information on children who did not receive i.v. antibiotics).

Blood tests at diagnosis

Diagnostic surgery

The types of surgery carried out post presentation and prior to start of i.v. antibiotics are given in Table 22. In simple cases, the types of surgery carried out most frequently were incision and drainage, and aspiration (60.7% and 55.7%, respectively), with arthroscopy third (9.8%). In complex cases, incision and drainage was the most common type of surgery (62.1%), followed by aspiration (24.1%) and debridement (17.2%) (see Table 22).

Some types of surgery were not carried out prior to i.v. treatment: amputation, compartment decompression, secondary closure, skin graft and other plastic surgery.

Diagnostic imaging

Radiography was widely used as a diagnostic tool (64.5% of children who received i.v. antibiotics). The proportion of radiographs that were abnormal was 38% for simple cases and 48% for complex cases.

The next most commonly used imaging method was ultrasonography (41% of simple cases and 31% of complex cases). Of those imaged, 74.1% detected an abnormality.

Magnetic resonance imaging was used in 19% of simple cases, with an abnormality being detected in 89%, and in 23% of complex cases, with a detection rate of 95%. Table 24 gives the numbers of children presenting with each bone/joint that had a diagnostic MRI. CT and bone scanning were rarely carried out, and in only OM cases. In all four cases, CT revealed abnormalities (Table 25).

Microbiology

Antibiotic treatment

Antibiotic regimes: overall

At initiation of i.v. treatment, over half of children (both types of case combined) were given one antibiotic (57.7%).

The timing of the switch from i.v. to oral antibiotics was different in simple cases and complex cases. In simple cases, switch occurred a median of 7 days after the start of i.v. therapy, whereas in complex cases this was a median of 13.5 days. Table 33 shows the percentages of children in whom switch dates were known who switched < 1 week, 2 weeks and > 2 weeks after commencement of i.v. treatment. Figure 1 shows a profile of time-to-switch data in the form of a Kaplan–Meier plot. This clearly shows that in complex cases the switch to oral antibiotics took place later, and that the time taken for i.v. therapy to be stopped or switched to oral therapy in half of children was 9 days in simple cases, compared with 21 days in complex cases (time point at which the curve reaches 0.5 on the y-axis).

TABLE 33 - Types of antibiotic regimes experienced

HF, higher formulary.

Higher formulary dose treatments recommended for bone/joint infections.

Antibiotic regime	Simple cases (n = 218)		Complex cases (n = 95)
	n (%)	Median (IQR), days	n (%)	Median (IQR), days
Number of different antibiotics given at initiation of i.v. treatment
1	131 (60.1)		53 (55.8)
≥ 2	72 (33)		38 (40)
No i.v. antibiotics given	12 (5.5)		3 (3.2)
Unknown	3 (1.4)		1 (1.1)
Timing of oral switch
Switch date known (weeks)	168 (77.1)	7 (5–13)	62 (65.3)	13 (7–24)
< 1	73 (43.5)	4 (3–5)	12 (19.4)	3.5 (2.5–5.5)
1–2	62 (36.9)	10 (8–13)	21 (33.9)	9 (8–10.5)
> 2	33 (19.6)	21 (16–30)	29 (46.8)	26 (20–41)
No switch date	50 (22.9)		33 (34.7)
Child stayed on i.v. antibiotics	35 (70)	15 (7–30)	29 (87.9)	25 (11–44)
No i.v. antibiotics	15 (30)		4 (12.1)
Total course of antibiotics (weeks)
< 6	139 (63.8)		49 (51.6)
≥ 6	71 (32.6)		43 (45.3)
Had no antibiotics/unknown	8 (3.7)		3 (3.2)
Dose
First dose HFa	88 (40.4)		28 (29.5)
HFa throughout	30 (13.8)		9 (9.5)
No HFa doses given	53 (24.3)		23 (24.2)
Combination	124 (56.9)		59 (62.1)
Unknown	11 (5)		4 (4.2)
Total number of antibiotics prescribed
i.v.
0	15 (6.9)		4 (4.2)
1	88 (40.4)		26 (27.4)
2	81 (37.2)		31 (32.6)
> 2	34 (15.6)		34 (35.8)
Oral
0	41 (18.8)		28 (29.5)
1	129 (59.2)		41 (43.2)
2	43 (19.7)		20 (21.1)
> 2	5 (2.3)		6 (6.3)

FIGURE 1.

Kaplan–Meier plot showing time to switch.

In the majority of cases with no known switch date, there was no record of any oral antibiotics having been prescribed following treatment with i.v. antibiotics.

The proportion of children who received antibiotic treatment for < 6 weeks in total was 63.8% for complex cases, compared with 51.6% for complex cases. Figure 2 shows that, among simple cases, there was little difference in time to switch to oral antibiotics between those with OM and those with SA. Among complex cases, switch timing was similar in the OM and SA subgroups when switch took place before 2 weeks, but among those who received i.v. antibiotics for at least 2 weeks, children with OM received i.v. antibiotics for longer, and were switched to oral antibiotics later, than children with SA.

FIGURE 2.

Kaplan–Meier plot showing time to switch (by SA/OM).

For each antibiotic prescribed, a daily dose per kilogram body weight was calculated. Each dose was classified as ‘higher formulary’ (recommended intensive dosing as recommended by the current bone and joint infection guidelines on the treatment of severe infections) or ‘not higher formulary’ [following standard British National Formulary for Children (BNFC) dosing guidelines for antibiotics62]. The proportion of children who received a higher formulary initial dose of antibiotics was 40.4% in simple cases, compared with 29.5% in complex cases. Most children received both higher formulary doses and not higher formulary doses during their treatment (57% for simple cases and 62% for complex cases). A total of 13.8% of simple cases received only higher formulary doses throughout their treatment, compared with 9.5% of complex cases (see Table 33).

The total number of antibiotics prescribed for each route was calculated for each child. So, for example, in the case of a child for whom there were two records of oral flucloxacillin and one record of co-amoxiclav, the total number of oral antibiotics recorded would be two. Note that this calculation is limited to instances of the antibiotics flucloxacillin, clindamycin, cefuroxime, ceftriaxone, amoxicillin, co-amoxiclav, rifampicin, vancomycin, fusidic acid, benzylpenicillin, teicoplanin, and ‘other’. All ‘other’ antibiotics were combined as one type for the sake of parsimony. In simple cases, a single i.v. antibiotic and/or a single oral antibiotic was the most common treatment strategy employed, whereas for complex cases there were cases of one, two or more than two i.v. antibiotics being used together with a single oral antibiotic type (see Table 33).

Adherence to discharge antibiotics

Almost three-quarters of children were followed up (71.6% of simple cases and 80.4% of complex cases, no difference between SA and OM). Of these, almost all reported that antibiotics were completed as planned (92.4% of simple cases and 92.1% of complex cases) (Tables 47 and 48 give further details of the reasons for non-adherence in the nine children who did not complete the course of prescribed antibiotics).

Complications

A total of 8.3% of simple cases and 16.8% of complex cases experienced at least one complication. Table 49 lists the numbers of types of complication by action taken. The types listed were options given for data entry; as Table 49 shows, many were not experienced (dislocation, subluxation, pathological complication, fracture and sinus). The most selected category was ‘other’: text entered to give more detail about these complications is given in Tables 50 and 51.

Ongoing symptoms

Of the children who were followed up, 27 (17.2%) simple cases had at least one ongoing symptom; 18 of these were still in pain and other symptoms were less common. Eighteen (24%) complex cases had at least one ongoing symptom. Types of symptoms were more diverse in this group: eight children were still in pain, eight had joint stiffness and five had a deformity (Table 52).

Treatment failure

Treatment failure is defined as any of the following complications or ongoing symptoms at follow-up:

• fever

• recurrence of infection

• dislocation

• subluxation

• pathological fracture

• development of a draining sinus

• other (pain, stiffness, mobility problems, swelling, numbness, increase in CRP compared with discharge).

And/or any of the following actions arising from a complication:

• readmission to hospital

• surgery

• change of antibiotics.

Under this definition, 32 (14.7%) simple cases experienced a treatment failure, compared with 24 (25%) complex cases (see Tables 53–56 for associations between treatment failure and diagnosis, switch timing, length of time on antibiotics and surgery during hospital stay). Observed associations do not necessarily imply causal links that explain likelihood of treatment failure.

Diagnosis group

Among simple cases, treatment failure was half as likely in children with SA as in children with OM (9.3% vs. 19.3%, respectively), whereas among complex cases treatment failure was more likely in children with SA than in those with OM (28.6% vs. 21.4%, respectively) (Table 53).

Switch timing

In simple cases, treatment failure was less likely the earlier switch occurred. Of the 73 simple cases in whom oral switch was instituted very early (< 1 week), seven (9.6%) experienced a treatment failure, compared with 10/62 (16.1%) of those in whom oral switch was early (between 1 and 2 weeks) and 6/33 (18.2%) of those in whom switch was standard (after 2 weeks of i.v. antibiotics). The risk of treatment failure can also be presented as odds ratios, which show that the risk of treatment failure following very early switch is half that of associated with switch after 2 weeks. Among complex cases, the risk of complications was highest in those in whom oral switch occurred after 2 weeks, at 37.9%, and lower in those undergoing switch very early (25%) or after 1–2 weeks of i.v. antibiotics (19%) (Table 54).

Length of time on antibiotics

In simple cases, treatment failure was equally likely for children who had antibiotics for < 6 weeks as for those who had courses for > 6 weeks. Among complex cases, however, children given antibiotics for < 6 weeks were much less likely to experience treatment failure than those given antibiotics for > 6 weeks (14.3% vs. 37%, respectively) (Table 55).

Surgery during hospital stay

Table 56 shows the numbers of children who experienced treatment failure grouped according to whether or not surgery was carried out during their hospital stay. There was no indication of an association between treatment failure and surgery in either simple or complex cases.

Discussion

This multicentre national UK service evaluation has defined the current case load, disease spectrum and clinical practice in the diagnosis and treatment of OM/SA in secondary and tertiary UK care to provide information regarding the feasibility of a future RCT design.

Blood tests

In order to inform diagnostic decisions, we organised our data collection to record the details of blood tests both at presentation and on the day of or the day preceding the diagnosis. It was perhaps surprising that, in our cohort, only 105 (48.2%) of the children with simple disease were recorded as having blood tests taken at presentation (see Table 14). The fact that these data also show only 33.7% of complex case children to have had a blood test at presentation is surprising, as this figure might have been expected to be higher than for simple cases. At first glance this suggests that our database question may not have been worded with enough clarity for the research staff to extract the correct information from the clinical records, or that not all blood tests carried out were recorded. However, we also collected information on the blood tests performed on the day of or the day immediately preceding the initiation of i.v. antibiotics (see Table 19), the time point at which, for most cases, there is an assumption that the diagnosis was made. It is also likely that clinicians performed blood tests at any point from presentation (with or without definitive symptoms suggestive of the diagnosis) or at any point up to when the diagnosis was actually made. As clinicians try to avoid repeated blood tests in children unless there is clinical change or clinical uncertainty, it is probable that the blood test data are valid but that in some cases only the results of blood tests carried out at presentation are available, whereas, in others, only the results of blood tests just before initiation of antibiotics (i.e. just before diagnosis) are available and in yet other cases results at both time points are available.

We analysed whether WBC/neutrophil counts, CRP and ESR were measured at presentation or before diagnosis as well as the values of these parameters to inform decisions regarding potential entry to a clinical trial (i.e. the usefulness in making the diagnosis of bone and/or joint infection). We also tried to establish at which point the CRP reached the maximum (CRP_max) recorded value to inform both diagnostic and subsequent monitoring decisions. These data need to be interpreted in light of the fact that the CRP_max recorded may have occurred before or after the initiation of antibiotics.

Of the currently available and routinely requested non-specific blood test biomarkers, the diagnosis of OM or SA was usually based on the CRP measurement alone (although a low CRP does not exclude OM or SA). At presentation, CRP was higher in complex disease than simple disease (simple cases: median 32 mg/l, IQR 11.8–66.7 mg/l; complex cases: median 54 mg/l, IQR 24–141 mg/l), which was not true for ESR (simple cases: median 32 mm/hour, IQR 20.5–70 mm/hour; complex cases: median 28 mm/hour, IQR 15–55 mm/hour) (see Table 15). At this time point, neither CRP nor ESR was able to distinguish between SA and OM (CRP: SA median 32 mg/l, IQR 16.2–65 mg/l; OM median 32.7 mg/l, IQR 10–68 mg/l; ESR: SA median 29 mm/hour, IQR 14–70.5 mm/hour; OM median 35.5 mm/hour, IQR 26–68.5 mm/hour) (see Table 16). As expected, the number of ESR results available for analysis was only around half of the number of CRP results (an ESR result was available in 56/218 simple cases and 9/95 complex cases, whereas a CRP measurement was available in 99/218 simple cases and 31/95 complex case, (see Table 15). This was probably due to both the perceived lack of need (many paediatric clinicians are less likely to request ESR in current practice) and the related fact that ESR is a technically harder test to perform than CRP because of the blood volume required and laboratory processes. A similar analysis is available for the blood tests performed on the day of or preceding the day of diagnosis. At this time point, CRP was also higher in complex disease than in simple disease (simple cases: median 38.3 mg/l, IQR 20–74 mg/l; complex cases: median 86.7 mg/l, IQR 29–161 mg/l), which was, again, not so for ESR (simple cases: median 46 mm/hour, IQR 25–78 mm/hour; complex cases: median 40 mm/hour, IQR 21–60) mm/hour) (see Table 20). Neither CRP nor ESR was able to distinguish between SA and OM (CRP: SA median 35.5 mg/l, IQR 21.2–72 mg/l; OM median 39.5 mg/l, IQR 14–81 mg/l; ESR: SA median 46 mm/hour, IQR 23–81 mm/hour; OM median 47 mm/hour, IQR 26–76 mm/hour) (see Table 21). Again, the number of ESR results available for analysis was half the number of CRP results (an ESR result was available in 83/218 simple cases and 17/95 complex cases, whereas a CRP result was available in 137/218 simple cases and 48/95 complex cases) (see Table 20).

Overall, CRP reached its highest recorded value at the start of the illness in simple disease and a few days after presentation in complex disease [in simple disease CRP_max occurred on median day 0 (IQR day 0–6) and in complex disease on median day 4 (IQR day 1–8)] (see Table 17).

The total WBC or specific neutrophil counts at both presentation or just prior to initiation antibiotics were not shown to be of clinical use in determining whether to suspect or confirm SA or OM. Although the normal ranges of total WBC and neutrophil counts vary at different ages, there was overlap for both parameters between the normal range and WBC/neutrophil values recorded for children with simple or complex OM and SA (see Tables 15, 16, 20 and 21). As for many other serious infections in children, a raised WBC and/or neutrophil count can indicate bacterial infection, but normal results do not preclude such a diagnosis and must not be used to rule out OM or SA.

Imaging

The use of imaging performed prior to i.v. antibiotics (prior to diagnosis of bone or joint infection) was determined. Overall, a radiological diagnosis was obtained for 51.8% of simple cases (at a median of 1 day post presentation, IQR 0–3) and 52.6% of complex cases (median of 1 day post presentation, IQR 0–4 days) (see Table 17), although, as can be seen from the detailed discussion in the next paragraph, these data may be skewed by soft-tissue abnormalities reported in formal X-ray reports. Regardless of the caution needed in interpreting this result overall, MRI has been shown to offer very high pick-up rates of abnormalities compared with X-ray or ultrasonography, and has almost completely replaced technetium bone scanning in current UK practice (see Tables 25 and 26). In detail, X-rays were commonly used (66% of simple cases of OM or SA combined, 62% of complex cases), ultrasonography was used in 41% of simple cases and 31% of complex cases and MRI was used in 19% of simple and 21% of complex cases. CT and technetium bone scanning were rarely used (1% of children with simple or complex disease underwent CT, whereas 2% of children with simple disease but none with complex disease underwent bone scanning) (see Table 25).

X-ray and ultrasonography were used with equal frequency in simple cases of OM and SA (see Table 26). Around two-thirds of children with simple or complex SA and OM underwent radiography before initiation of i.v. antibiotics (see Table 18). Of the children with simple SA, 66 (63%) had an X-ray, of which 24 (36%) were abnormal. Of the children with simple OM, 66 (68%) had an X-ray, of which 26 (39%) were reported as abnormal. Among those with complex disease, X-rays taken prior to antibiotics were abnormal in 8 of 21 (38%) of those with SA, compared with 18 of the 33 (55%) of those with OM. Because of the nature of the data collection in the study, these data must be interpreted with caution as it is likely that abnormalities recorded in the database included soft tissue changes visible (and reported by radiologists) and not just bony abnormalities that would not be expected to be present early in the OM disease process. Ultrasonography was carried out in 53 (51%) simple SA cases and was reported abnormal to be abnormal in 43 of these (81%). Eleven children with complex SA underwent ultrasonography (33% cases), and was reported to be abnormal in nine (82%). In complex OM there were 16 scans (30% cases), of which 11 (69%) were reported as abnormal.

Magnetic resonance imaging was used more often in children with OM cases than in those with SA and picked up a large number of abnormalities (see Table 26). Twenty-nine (30%) children with simple OM underwent MRI, in 26 (90%) of whom the results were reported as abnormal, and 16 (30%) children with complex OM underwent MRI, in all of whom the results were reported to be abnormal. MRI was also very successful in the diagnosis of SA. MRI was carried out in nine (9%) children with simple SA, with abnormal results reported in eight (89%), and in four (12%) children with complex SA. with abnormal results reported in three (75%).

Among children with OM, whether simple or complex, MRI revealed abnormalities in many different bones, whereas in children with SA abnormalities on MIR were seen only in large joints (see Table 24). The ability of MRI to detect a wide range of abnormalities explains the low frequency with which technetium bone scanning was performed over the course of the evaluation period. Bone scanning was performed in no children with complex SA or OM, in only one child with simple SA (in whom the result was abnormal) and only in three children (3%) with simple OM, of whom two (67%) showed an abnormality. CT was not used at all in children with SA, and in only three children (3%) with simple OM (100% abnormal) and one child (2%) with complex OM, in whom the results were was also abnormal (see Table 26).

Surgery

Surgery is carried out before starting i.v. antibiotics in children with clinically severe SA or OM either to prevent destructive outcomes or to obtain surgical specimens to aid a microbiological diagnosis, or where there is diagnostic dilemma.

Diagnostic surgery prior to i.v. antibiotic therapy was carried out in a higher proportion of SA cases than of OM cases [39/107 (36.4%) simple and 15/35 (42.9%) complex cases of SA, compared with 21/109 (19.3%) simple and 14/56 (25%) complex cases of OM]. On average, surgery was carried out earlier in children with SA than in those with OM. Among those with SA, the first surgery was carried out a median of 1 day after presentation, with 75% of first surgical procedures being carried by day 2 in simple cases and by day 5 in complex cases. Among children with OM, the first surgery was carried out a median of 1 day after presentation in simple cases, with 75% of first surgical procedures performed by day 10, and a median of 3 days after presentation in complex cases, with 75% of first surgical procedures performed by day 8 (see Table 18).

The types of surgery most often carried out before i.v. antibiotics were started were, in simple cases (both OM and SA), ‘incision and drainage’ (performed in 60.7%) and ‘aspiration’ (in 55.7%), followed by arthroscopy (9.8%), and, in complex cases, incision and drainage (62.1%), then aspiration (24.1%) and, third, debridement (17.2%) (see Table 22).

Surgery was carried out more often in children with OM than in those with SA (63.6% vs. 37.5%, respectively, in simple cases and 25.9% and 15.5%, respectively, in complex cases). From discussion at the investigator meeting (see Chapter 5), there remains the possibility that, although the database has been populated directly from terms used in the clinical records, there may be overlap between the different technical terms used by orthopaedic surgeons for ‘incision and drainage’ and ‘aspiration’. At database design we had intended ‘aspiration’ to record needle aspiration (expected to be mainly of infected joints in suspected SA) and ‘incision and drainage’ to represent a more formal, open surgical procedure (expected to be carried out in cases of both OM and SA). We found that incision and drainage was performed with approximately equal frequency in both groups (simple SA 25%, simple OM 33%, complex SA 13.4%, complex OM 9.3%) but there were also no reported differences in aspiration rates between SA and OM (simple cases SA 23.1%, OM 30.3%; complex cases SA 5.2%, OM 3.7%).

Surgical treatment after the initiation of antibiotics

A total of 21.6% of children with simple disease went on to have surgery after the start of i.v. antibiotics, compared with 41.1% of children with complex disease. In both groups, OM and SA cases were equally represented (see Table 40).

The most common procedure carried out in simple cases was aspiration (7.3%); in complex cases it was incision and drainage (16.8%), and this was often a wash-out in conjunction with another procedure. In 11.6% of all complex cases, debridement was performed at least once (see Table 40).

Conventional microbiological tests

As for any invasive infections in children or adults, clinical best practice in OM and SA is to take samples for microbiological diagnosis prior to starting empirical antibiotics. The aim of tests is to make a microbiological diagnosis in order to determine further, more specific antibiotic therapy and, in the case of infection with some specific pathogens such as PVL-positive Staphylococcus, to consider more complex or prolonged treatment regimens. Routinely available NHS microbiological tests are reported here, with a summary of the molecular data from a new targeted PCR test reported in Chapter 3.

In only 144 out of 218 (66.1%) children with simple OM or SA, and 59 out of 95 (62.1%) children with complex OM or SA, was at least one diagnostic test performed on blood or tissue samples (bone, joint aspirate or periosteal pus) (see Table 27). Blood culture was performed in only 62% of simple cases and 54.3% of complex cases. Blood culture reporting rates did not differ between secondary and tertiary centres (see Table 30).

A blood, tissue or synovial fluid sample tested positive for at least one micro-organism in 38.2% of children with simple disease, compared with 52.5% of complex cases. Among those children with OM, microbiology tests were carried out in 58%, and at least one micro-organism was detected in 49% of cases tested. The proportion of children with SA tested was 73%, and a micro-organism was found in 36% (see Table 27).

S. aureus was found in 20.1% of blood, tissue or synovial fluid samples taken from simple cases and in 32.2% of samples from complex cases; other pathogens were far less commonly detected by routinely available tests (see Table 27). PVL-positive S. aureus and MRSA were detected with similar frequency in simple and complex cases regardless of sample type [PVL-positive S. aureus: 2/29 (6.9%) simple cases and 1/19 (5.3%) complex cases; MRSA: 1/29 (3.4%) simple cases and 1/19 (5.3%) complex cases]. In only one case (complex) was erythromycin-resistant S. aureus reported (see Table 27). GAS, K. kingae and GBS were not commonly detected by conventional microbiology (see Table 27).

Medical treatment: intravenous course length and time before oral switch

All children in the study spent time in hospital. Most children with SA or OM went straight to hospital [median time from presentation to hospital referral 0 days (IQR 0–3 days) in simple cases and 1 day (IQR 0–4 days) in complex cases] (see Tables 17 and 18).

Not all children received i.v. antibiotics, although most did [104/107 (97.2%) simple SA cases, 33/95 (94.3%) complex SA cases, 97/109 (97%) simple OM cases and 54/56 (96.4%) complex OM cases]. In 15 (6.9%) simple cases and four (4.2%) complex cases, either no i.v. antibiotic treatment was given or the data are missing (see Tables 17 and 18). Among those who were treated with i.v. antibiotics, treatment was initiated, on average, a little earlier in children with SA [simple cases, median 1 day (IQR 0–3 days); complex cases, median 2 days (IQR 0–4 days)] than in those with OM [simple cases, median 2 days (IQR 1–5 days); complex cases, median 2 days (IQR 1–5 days)] (see Table 18).

Although there was no difference in overall duration of antibiotic treatment between simple and complex SA cases or between simple and complex OM cases, duration of i.v. therapy was longer in children with complex disease, whether SA or OM, than in those with simple disease (see Table 32):

• Among the 103 children with simple SA treated with antibiotics, the median duration of total treatment was 30 (IQR 18–41) days, with a median duration of i.v. therapy of 7 (IQR 4–13) days.

• Among the 34 children with complex SA treated with antibiotics, the median duration of total treatment was 28.5 (IQR 15–46) days, with a median duration of i.v. therapy of 14 (IQR 7–24) days.

• Among the 105 children with simple OM treated with antibiotics, the median duration of total treatment was 40 (IQR 21–48) days, with a median duration of i.v. therapy of 9 (IQR 6–16) days.

• Among the 54 children with complex OM treated with antibiotics, the median duration of total treatment was 42 (IQR 26–57) days, with a median duration of i.v. therapy of 18.5 (IQR 8–42) days.

These data suggest that in any potential clinical trial of antibiotics in simple OM or SA, although the duration of total may need to differ between OM and SA, it may be possible to plan for a similar length of initial i.v. therapy for both conditions.

The timing of switch from i.v. to oral antibiotics was related to disease complexity and type of disease: SA or OM. Intravenous antibiotics treatment was of shorter duration in children with simple disease than in those with complex disease (see Figures 1 and 2 and Tables 32 and 33). Among children in whom the switch date is known, a blood test at the time of the switch was carried out in only 37.9% of those with simple disease and 22.8% of those with complex disease (see Table 38).

When the switch date was recorded in the database, the median duration of i.v. therapy could be calculated:

• Among 107 simple cases of SA, the switch date was known in 85 (79.4%). Switch occurred after a median of 6 (IQR 4–11) days of i.v. antibiotic treatment.

• Among 109 simple cases of OM, the switch date was known in 82 (75.2%). Switch occurred after a median of 8.5 (IQR 5–14) days of i.v. antibiotic treatment.

• Among 35 complex cases of SA, the switch date was known in 23 (65.7%). Switch occurred after a median of 13.5 (IQR 7–21) days of i.v. antibiotic treatment.

• Among 56 complex cases of OM, the switch date was known in 35 (62.5%). Switch occurred after a median of 10.5 (IQR 7.5–26) days of i.v. antibiotic treatment.

As described above, the Kaplan–Meier plot of time to switch (by SA/OM; see Figure 2) shows that time to switch to oral antibiotics differed little between those with OM and SA in the simple cases group. Among complex cases, switch timing was similar in those with OM and SA when the switch took place before 2 weeks, but after this time point switch to oral antibiotics occurred later in those with OM than in those with SA.

Overall, 63.8% of those with simple disease (OM/SA) received antibiotic treatment for < 6 weeks in total, compared with 51.6% of those with complex disease (see Table 33). Simple OM and SA were most commonly treated with a single i.v. antibiotic and/or a single oral antibiotic, whereas in complex SA/OM cases more than two i.v. antibiotics were used, most often together with a single oral antibiotic type (see Table 33).

Medical treatment: initial empirical antibiotics

In order to design a randomised clinical trial, consensus needs to be reached regarding the initial empirical antibiotic to be started intravenously, and also regarding the post-switch oral antibiotic(s) to be used for the cases in which there is or is not a formal microbiology diagnosis. Although not investigated in this study, data from Finland suggest that children with culture-negative signs and symptoms of SA or OM can be treated in a similar fashion to children in whom the organism is identified in respect of type and duration of antimicrobial therapy. 63 The aim of the microbiology component of this study was to establish the most appropriate initial empirical therapies.

Beta-lactam antibiotics, either ‘simpler’ penicillin or second- or third-generation cephalosporins, are widely used in UK practice for the treatment of SA and OM. Among the 199 children with simple SA or OM who received i.v. antibiotics at any point during their hospital stay, the initial i.v. treatment was flucloxacillin in 85 (42.7%), clindamycin in 13 (6.5%), cefuroxime in 60 (30.2%), ceftriaxone in 51 (25.6%), benzylpenicillin in 16 (8%) and co-amoxiclav in 13 (6.5%) [24 (12.1%) children received other antibiotics] (see Table 35). Of the 91 children with complex SA or OM who received i.v. antibiotics at any point during their hospital stay, 38 (41.8%) received flucloxacillin, 10 (11%) received clindamycin, 17 (18.7%) received cefuroxime, 19 (20.9%) received ceftriaxone, 10 (11%) received benzylpenicillin and 10 (11%) received co-amoxiclav [with 19 (20.9%) children receiving other antibiotics] as part of their initial treatment (see Table 36).

Oral antibiotics were uncommon as part of initial therapy. Seventy-eight (35.8%) children with simple disease were prescribed at least one oral antibiotic. Flucloxacillin (34 cases, 43.6%), clindamycin (17 cases, 21.8%) and co-amoxiclav (18 cases, 23.1%) were all prescribed prior to discharge from hospital (see Table 35). Thirty-six (37.9%) children with complex disease were prescribed at least one oral antibiotic while in hospital, with the range of antimicrobials prescribed similar to those prescribed for simple disease, although ‘other’ antibiotics were used more often in children with complex disease than in those with simple disease (Tables 35 and 36).

Medical treatment: antibiotic at hospital discharge

Children with OM or SA were discharged on either i.v. or oral antibiotics. At discharge, 30 (28%) of 107 children with simple SA and 28 (25.7%) of 109 children simple OM remained on i.v. antibiotics. Around 60% [SA, 64 cases, (59.8%); OM, 65 cases (59.6%)] of children were taking only an oral antibiotic, and similar numbers of children with simple SA and OM either were on no antibiotic or the data are missing [SA, 13 children (12.1%); OM, 16 children (14.7%)]. Children with complex disease were more likely to remain on i.v. medication at discharge [of 95 children with complex SA/OM, 40 (42.1%) were on i.v. antibiotics, 42 (44.2%) were on oral antibiotics and 13 (13.7%) children were on none or this information was unknown] (see Table 45).

As expected, owing to the formulations available for outpatient or home-based therapy, the range of drugs prescribed was narrower than those used in hospital. Most children with simple disease were prescribed ceftriaxone for post-discharge i.v. therapy (41/59; 69.5%), although a few remained on flucloxacillin (4/59; 6.8%), clindamycin (2/59; 3.4%), cefuroxime (1/59; 1.7%) or other drugs (9/59, 15.3%). Children with complex disease were also treated after discharge mostly with ceftriaxone (31/40; 77.5%), but a few were treated with flucloxacillin (2/40; 5%), clindamycin (3/40; 7.5%) or other drugs (5/40; 12.5%) (see Table 46). Note: for 10% of cases known to have been discharged on i.v. antibiotics, the antibiotic type was not recorded, and children could be discharged on more than one antibiotic.

Of the oral drugs prescribed at discharge, co-amoxiclav was used in approximately one-third of cases [58/156 (36.5%) simple cases, 17/58 (28.8%) complex cases] and flucloxacillin in about one-quarter of cases [46/156 (28.9%) simple cases, 15/58 (25.4%) complex cases]. Oral clindamycin was used more frequently in complex than in simple cases [17/156 (10.7%) simple cases, 17/58 (28.8%) complex cases]. Cefalexin was used in simple SA/OM but infrequently in complex disease [20/156 (12.6%) simple cases, 2/58 (3.4%) complex cases] (see Table 46).

Medical treatment: dose of intravenous and oral antibiotics prescribed

Adequate dosing is a major concern of modern antimicrobial stewardship to ensure clinical cure with minimal risk of intrapatient and in-hospital interpatient antimicrobial resistance. 64 Recent legislative changes to clinical trial legislation65 ensure that newly available drugs are properly tested in children to establish dosing regimens and efficacy. However, almost all the widely used paediatric doses of antibiotics have been included in paediatric formularies, having been established by opinion and practice rather than formal clinical trial. With much variation in international practice, the UK BNFC62 has recently begun to revise the recommended doses of commonly used antibiotics, for example in 2014 doubling the recommended age band doses of oral amoxicillin. In addition, the UK BNFC is set out so that minimum and maximum doses for weight or age are specified. Many specialists in paediatric infection and paediatric microbiology, and National Institute for Health and Care Excellence guidelines where they exist,66–68 consider the high quoted doses to be the correct doses for serious (in-hospital) infections and that the low quoted doses are suboptimal. In order to inform the dose strategy for a future clinical trial, it was important to assess the antibiotic doses used in current practice in the context of overall reported clinical outcomes.

Overall, only 40.4% of simple cases received a higher formulary initial dose of antibiotics, compared with 29.5% of complex cases. Only 30 (13.8%) simple cases and nine (9.5%) complex cases received the higher formulary dose of antibiotic treatment throughout (see Table 33).

Specifically, for the most commonly prescribed i.v. empirical antimicrobials in hospital:

• For simple disease (see Table 35), flucloxacillin, clindamycin and cefuroxime were prescribed at or near to the recommended formulary higher doses [flucloxacillin: median 188.1 (IQR 100.8–200) mg/kg/24 hours, formulary higher dose 200 mg/kg/24 hours; clindamycin: median 39.9 (IQR 18.5–40.6) mg/kg/24 hours, formulary higher dose 40 mg/kg/24 hours; cefuroxime: median 151.5 (IQR 83.7–150.) mg/kg/24 hours, formulary higher dose 150 mg/kg/24 hours]. However, ceftriaxone was generally prescribed at the lower formulary dose [median 50.3 (IQR 50–79.3) mg/kg/24 hours, formulary higher dose 80 mg/kg/24 hours].

• For complex disease (see Table 36), however, prescribing trends suggest that inadequate dosing (less than the higher formulary dose) was more common. Although ceftriaxone was generally prescribed at the higher formulary dose when it was used (as initial treatment in 20.9% of cases, and in 40.7% of complex cases overall) [median 77.7 (IQR 49–8) mg/kg/24 hours, formulary higher dose 80 mg/kg/24 hours, EMA-approved maximum dose 100 mg/kg/24 hours], flucloxacillin, clindamycin and cefuroxime were all prescribed at less than the formulary higher doses [flucloxacillin: median 142.6 (IQR 98–198.5) mg/kg/24 hours, formulary higher dose 200 mg/kg/24 hours; clindamycin: median 33.3 (IQR 24.1–39.2) mg/kg/24 hours, formulary higher dose 40 mg/kg/24 hours; cefuroxime: median 128 (IQR 83.3–150) mg/kg/24 hours, formulary higher dose 150 mg/kg/24 hours].

Doses of i.v. and oral antimicrobials prescribed following discharge were also generally lower than higher formulary doses. The i.v. empirical antimicrobial most commonly prescribed at discharge (for use out of hospital) (see Table 46), in both simple and complex cases, was ceftriaxone, because of its once-daily regimen (prescribed in 69.5% of children with simple disease receiving i.v. home-based therapy and 77.5% of children with complex disease). At discharge, ceftriaxone was generally prescribed at the lower formulary dose in simple cases [median 50.7 (IQR 49.3–79.4) mg/kg/24 hours, formulary higher dose 80 mg/kg/24 hours], although in complex disease the dose was slightly higher but still not at the higher formulary dose [median 65.4 (IQR 49–80.3) mg/kg/24 hours]. When used out of hospital, i.v. flucloxacillin and clindamycin for both simple and complex disease were generally prescribed at below the lower formulary recommended doses (flucloxacillin: simple disease, median 98.1 mg/kg/24 hours; complex disease, median 35.8 mg/kg/24 hours; clindamycin: simple disease, median 21.6 mg/kg/24 hours; complex disease, median 4 mg/kg/24 hours) (see Table 46).

In the case of orally prescribed out-of-hospital antibiotic treatments, comparison with formulary doses is more complicated as a result of the age band variations in the current BNFC. 62 For reference, the recommended dose ranges for co-amoxiclav (expressed as mg/kg/24 hours of amoxicillin component) are as follows:

• age 1 month to 6 years: low dose 18.75 mg/kg/24 hours; high dose 37.5 mg/kg/24 hours

• age 6–12 years: low dose 22.5 mg/kg/24 hours; high dose 45 mg/kg/24 hours

• age 12–18 years: low dose 15 mg/kg/24 hours; high dose 30 mg/kg/24 hours.

For flucloxacillin, the recommended dose ranges are as follows:

• age 1 month to 2 years: low dose 31 mg/kg/24 hours; high dose 62 mg/kg/24 hours

• age 2–10 years: low dose 43.3 mg/kg/24 hours; high dose 46.5 mg/kg/24 hours

• age 2–18 years: low dose 25 mg/kg/24 hours; high dose 50 mg/kg/24 hours.

For all age groups, the recommended oral dose of clindamycin is 12 mg/kg/24 hours (low dose) to 24 mg/kg/24 hours (high dose) and of cephalexin is 25 mg/kg/24 hours (low dose) to 50 mg/kg/24 hours (high dose).

In most children with simple disease, the prescribed doses of the four most common oral antibiotics were nearer to the lower end of the dose range (see Table 46):

• co-amoxiclav: median 30.4 mg/kg/24 hours (IQR 13.8–41.8 mg/kg/24 hours)

• flucloxacillin: median 37.5 mg/kg/24 hours (IQR 14.2–82.3 mg/kg/24 hours)

• clindamycin: median 18.1 mg/kg/24 hours (IQR 6.1–23.6 mg/kg/24 hours)

• cefalexin: median 35 mg/kg/24 hours (IQR 24.5–91.1 mg/kg/24 hours).

Perhaps worryingly, children with complex disease who were prescribed co-amoxiclav or clindamycin (respectively 28.8% and 28.8% of children with complex disease given oral antibiotics) appear to have received lower doses of these drugs than children with simple OM or SA. Among children with complex disease prescribed oral antibiotics at discharge, the median dose of co-amoxiclav was 23.1 mg/kg/24 hours (IQR 12.1–47.6 mg/kg/24 hours), the median dose of flucloxacillin was 41.7 mg/kg/24 hours (IQR 36.4–54.1 mg/kg/24 hours) and the median dose of clindamycin was 15.1 mg/kg/24 hours (IQR 6–23.8 mg/kg/24 hours) (see Table 46).

Follow-up (3 months): completion of antimicrobial courses as planned at discharge

One of the key concerns in any clinical trial of oral therapies, and of antibiotics in particular, is the course completion rate. Although not all children with OM or SA were followed up (the reported follow-up rates were 72.9% of children with simple SA, 72.5% of children with simple OM, 77.1% of children with complex SA and 80.4% of children with complex OM), course completion rates among those who were followed up were very high (92.4% among those with simple disease and 92.1% among those with complex disease) (see Table 47). The reported reasons for non-completion of therapy were reassuring for future trial planning, when, to study clinical outcomes in relation to length of therapy, it will be important to know that prescribed medication is being given as planned. Among children with simple disease, there were only four for whom failure to complete treatment was reported, with the reported reasons being unpalatable antibiotic, rash, symptom resolution and refusal to take the medication.

Hospital stay

Overall, the median length of hospital stay among children with simple disease, whether OM or SA, was 7 days (IQR 4–12 days), with 50% of children staying for ≥ 6 days but no longer than 12 days. In complex cases the median length of stay in hospital overall was 9 days (IQR 6–18 days), with 50% of children staying for ≥ 6 days but no longer than 18 days (see Table 43).

Children with OM spent longer in hospital than those with SA, and this was true for both those with simple disease and those with complex disease (see Table 44). The median length of stay for simple SA was 6 days (IQR 4–11 days), compared with 9 days (IQR 5–13 days) for simple OM. The median length of stay for complex SA was 8 days (IQR 6–15 days), compared with 9 days (IQR 6–18.5 days) for complex OM. In the case of children with simple SA or OM, the median length of hospital stay was similar to the time on i.v. therapy prior to oral switch (oral switch occurring after a median of 6 and 8.5 days, respectively), but children with complex SA or OM were more likely to still be on i.v. therapy at the time of hospital discharge (oral switch occurring after a median of 13.5 and 10.5 days, respectively).

Younger (aged < 1 year) and older (aged 6–16 years) children were more likely to stay in hospital longer than the 1–6 years age group, which is probably a reflection of the difficulties of imposing bed rest on the younger mobile children than on babies and older children. Among those with simple disease, the median length of stay was 7 days (IQR 5–12 days) for those aged < 1 year and 10 days (IQR 6–15 days) for those aged 6–16 years but only 6 days (IQR 3.5–10.5) for those aged 1–6 years. Among those with complex disease, the median length of stay was 16 days (IQR 9–27 days) for those aged < 1 year and 10 days (IQR 6–16 days) for those aged 6–16 years, but only 8 days (IQR 5–14 days) for those aged 1–6 years (see Table 43).

The finding that children treated in secondary care centres spent slightly longer in hospital than those treated in tertiary centres (see Table 43) is difficult to explain fully, but may be a reflection of the fact that some children are transferred back to their secondary care centre prior to final discharge from hospital and this may incur delays overall prior to discharge home.

Follow-up (3 months): complications of the disease and therapies, and treatment failure

Complications and ongoing symptoms were recorded as separate items during data collection to try to capture all post-treatment events and problems that may need to be included as future clinical trial outcomes.

A total of 8.3% of children with simple disease and 16.8% of those with complex disease experienced at least one complication (see Table 49). In only three (of 218) children with simple disease and four (of 95) children with complex disease was recurrence of infection requiring medical or surgical intervention reported. Other complications were reported only once and were related to the disease (clinical outcomes, e.g. stiffness or pain), antibiotic therapy (e.g. possible allergy or drug reaction) or mode of giving the drug (e.g. phlebitis) (see Table 50). The complications of complex disease were generally more severe than those reported for simple disease, including acute kidney injury, Hickman line sepsis, pneumococcal meningitis and tracheostomy (see Table 50).

Twenty-seven (17.2%) children with simple disease and 18 (24%) with complex disease had at least one ongoing symptom. Eighteen of 157 (11.5%) children with simple SA or OM and 8 of 76 (10.5%) children with complex SA or OM followed up were still in pain. Six (3.8%) children with simple disease and eight (10.5%) with complex disease reported joint stiffness, and four (2.5%) children with simple disease and three (3.9%) with complex disease reported immobility of some kind. Other symptoms were less common and no fractures were reported (see Table 52).

To capture all possible cases of failure of initial empirical antibiotic and/or surgical therapy, necessitating further treatment, we used a broad definition of treatment failure (including fever, recurrence of infection including increase in CRP or development of draining sinus), dislocation, subluxation, pathological fracture, pain, mobility problems, swelling, numbness, hospital readmission, surgery and change of antibiotics. Using these criteria, 32 of 218 (14.7%) children with simple disease experienced a treatment failure, compared with 24 of 95 (25%) children with complex disease (see Table 53). Among those with simple disease, treatment failure was half as likely in those with SA (9.3%) as in those with OM (19.3%), whereas among those with complex disease treatment failure was more likely in those with SA (28.6%) than in those with OM (21.4%) (see Table 53).

These data demonstrate that treatment failure is less likely in children with simple disease in whom switch to oral antibiotics occurs early, suggesting that those with complex disease are more likely to fail initial therapy or require additional interventions, which is very important for the design of a RCT of reduced duration of i.v. and/or total antibiotic therapy. Of the 73 children with simple disease in whom oral switch occurred very early (< 1 week), 9.6% experienced a treatment failure, compared with 16.1% of those in whom switch occurred early (between 1 and 2 weeks) and 18.2% of those in whom switch took place after 2 weeks or more of i.v. antibiotics. Among children with complex disease, switching to oral antibiotics after 2 weeks or more was also associated with a higher risk of complications (37.9%) than very early switch (25%) or switch between 1 and 2 weeks (19%) (see Table 54).

Regarding the total length of therapy, among those with simple SA or OM, treatment failure was equally likely among those treated with antibiotics for < 6 weeks and those treated for > 6 weeks. Among those with complex disease, children given antibiotics for < 6 weeks were much less likely to experience treatment failure than those given antibiotics for > 6 weeks (14.3% compared with 37%), although this is likely to be a reflection of disease severity itself and not the treatments given (see Table 55).

Chapter 8 presents a detailed discussion of how the service evaluation data impact, with other study components, on potential RCT design.

Inclusion criteria

1. Children between the ages of 1 month and 16 years with a clinical diagnosis of OM or SA.

2. Written informed consent of participant or parent/legal guardian, and assent where appropriate for:

   • PCR to be done on routine samples taken for culture and microscopy

   • throat swab to be taken for PCR and culture and microscopy

   • additional 5-ml blood sample to be taken and stored for future analysis.

Exclusion criteria

Patients for whom informed consent is not obtained (parents/patients could decline consent for certain procedures, e.g. additional blood aliquot, but still give consent for others, such as PCR on routine samples).

Sample handling and preparation

Samples from routinely collected blood and tissue cultures were taken in the local microbiology laboratory and processed in accordance with the study standard operating procedure (see Report supplementary material 4) before storage and transfer to the central laboratory (Southampton Public Health England South East Regional Laboratory) for PCR analysis.

All samples were taken for PCR after initial bacterial culture had been carried out. For blood cultures, an aliquot of sample was removed after at least 19 hours’ incubation (unless the sample had already been identified as positive) using a sterile needle and syringe, from which 1 ml was placed into a sterile nuclease-free skirted 2-ml tube before being capped and labelled and placed in storage at –80 °C until extraction and PCR was to be carried out. For liquid samples such as synovial fluid, pus and blood or joint washouts, 500 µl was taken from the original sample (where limited sample volume remained, 200 µl was used, thus allowing for repeat culture or further analysis using another method). Solid samples were divided and, where sample quantity allowed, a minimum of two aliquots of tissue or bone weighing around 10 mg each were taken and placed in to a sterile nuclease-free skirted 2-ml tube and stored at –80 °C. All samples were handled in a class 1 cabinet to prevent contamination.

For throat swabs, after swabbing, the cotton end was snipped off using sterile scissors into a flat-bottom 2-ml tube containing 1 ml of skimmed milk, tryptone, glucose and glycerine media, used to preserve bacteria while being frozen at –80 °C.

Polymerase chain reaction analysis

Control bacteria used in this study were sourced from Pro-Lab Diagnostics (Birkenhead, UK) and the UK National Culture Collection (Salisbury, UK). Bacterial deoxyribonucleic acid was extracted by standard methods and multiplex PCR used for analysis. Table 57 shows the multiplex panels used for all samples.

Table 57 shows the constituent elements to each of the multiplexes. Two separate targets were used to detect S. pneumoniae as the pneumolysin (ply) gene is also common to other closely related streptococcal species including Streptococcus mitis, Streptococcus oralis and Streptococcus pseudopneumoniae. The N-acetylmuramoyl-L-alanine amidase (lytA) gene was later chosen as it occurs only in S. pneumoniae.