Immunosuppression

All patients with a record of an allogeneic HSCT between 2006 and 2015 were identified in the Patient Register (n = 2147). This cohort and methods used for identification of cGVHD have been described previously [14 (link)]. Briefly, the following exclusion criteria were applied: absence of a haematological malignancy prior to the HSCT; reused proxy identification numbers; age <18 or >75; and death within 6 months post-HSCT (S1 Fig). cGVHD is commonly defined as occurring onwards of 6 months post-HSCT [16 (link), 17 (link)]. For patients who survived ≥6 months post-HSCT (index date), the 0–6-month follow-up period was therefore 6–12 months post-HSCT. End of follow-up was Dec 31, 2016 (Cancer Register and Patient Register) and Dec 31, 2017 (Prescribed Drug Register and Cause of Death Register). Patient register records were reviewed, and patients were classified as having non-, mild, or moderate-severe cGVHD based on timing and extent of treatments commonly used for cGVHD using criteria developed by the authors (Fig 1) [14 (link)].
Patients were classified as non-cGVHD if, after taper of post-HSCT GvHD prophylaxis immunosuppression, they received neither systemic corticosteroids nor systemic immunosuppressive treatment (including everolimus, cyclosporine, methotrexate, mycophenolate, sirolimus, and tacrolimus) during the entire observation period. In Sweden, GvHD prophylaxis post-HSCT is discontinued by 3 months for matched sibling donors, and by 5 months for matched unrelated donors.
Patients with low-level cGVHD require less intensive immunosuppressive treatment. Based on this rationale, mild cGVHD was defined as patients receiving either corticosteroids or immunosuppressants alone. The following four mild cGVHD groups were defined: 1) patients who received systemic corticosteroid treatment >3 months alone, 2) patients whose last date of systemic corticosteroid treatment ended <3 months before censoring, 3) patients whose last date of systemic corticosteroid treatment ended <6 months before death, and 4) patients who received immunosuppressive treatment only (Fig 1).
Treatment modalities are similar for patients with moderate and severe cGVHD, which meant that it was not possible to separate these two groups. Patients with moderate-severe cGVHD require more intensive treatment than those with mild cGVHD. Based on this rationale, the following three moderate-severe cGVHD groups were defined: 1) patients who received corticosteroid treatment (irrespective of duration) and immunosuppressive treatment, 2) patients who received corticosteroid treatment (irrespective of duration) and extracorporeal photopheresis (ECP), and 3) patients who only received ECP.
Patients were assigned retrospectively to respective groups based on treatment (or not) received.

All consecutive patients aged 3 months to 18 years old with a diagnosis of acute OM and/or SA according to the International Classification of Diseases, 9th Revision, Clinical Modification code were evaluated for inclusion. A case was defined by diagnosis of OM or SA on imaging, preferably magnetic resonance imaging (MRI, gold standard) for OM, or in alternative computed tomography (CT scan), Tc99 bone scintiscan, PET-TC scan, or ultrasound (US)/MRI for SA. Long bones were considered the typical site of infection for OM. The hips were considered a high-risk site for both OM and SA. Exclusion criteria were diagnosis of immunodeficiency or hemoglobinopathy or chronic granulomatous disease, immunosuppressive therapy, concomitant systemic bacterial infection, and ongoing antibiotic treatment on admission. Patients with complicated infections, not fully vaccinated, and/or with incomplete follow-up were excluded, as well as those with chronic osteomyelitis and Brodie's abscess.
The population was divided into two main groups, OM and SA. Each group was further divided into three groups: pre-intervention, post-intervention not following the guidelines (no GL), and post-intervention group with adherence to the guidelines (GL).
The following variables, selected a priori, were evaluated: age, sex, weight, fever, vaccination status, white blood cells, and neutrophil count, CRP, erythrocyte sedimentation rate (ESR), and procalcitonin (PCT) at onset, IV and oral antibiotic treatment with duration, diagnosis and imaging type, typical vs. atypical site, results of blood, pus, synovial fluid cultures, MRSA colonization status, Quantiferon results, PVL test positivity, treatment failure (defined as treatment escalation to broad spectrum antibiotics and/or need for surgery) and relapse at six months of follow-up. PCR tests for identification of K. kingae or other pathogens in case of culture-negative infections were not performed, as not included as standard of care at our facility.