This was a retrospective study of all IVDU who were hospitalized for treatment of infectious diseases and were evaluated on demand by the infectious disease service over six years, January 2001 through December 2006, in the University Hospital Basel in Switzerland, a 780-bed primary and tertiary care center with approximately 27,000 admissions annually.

One single investigator extracted the following data from hospital charts as well as from the separate medical charts of the infectious disease specialists who were consulted: Name; gender; date of birth; date of admission; duration of hospitalization; date of infectious disease specialist consultations; duration of illicit drug use and which substances used; participation in an opioid maintenance program; main diagnosis and co morbidities; body mass index; infectious diseases diagnosis; main pathogen; additional pathogens; antibiotics, doses, application forms and durations of treatment; complications (i.e. acute renal failure, adult respiratory distress syndrome, disseminated intravasal coagulation, septic shock); admission to intensive care unit (ICU) and days on ICU; fatalities and reasons of death; fever, c-reactive protein and leucocytes at time of admission and time of discharge; time to defeverescence; readmission and reason for readmission. The data was entered in a database (SPSS 14, SPSS Inc., Chicago US), controlled by a second investigator and checked for plausibility during analysis of the dataset by another two independent reviewers.

Treatment protocols were evaluated with respect to their appropriateness for treating the diagnosed infection. Where treatment programs deviated from accepted standards, the reason for the use of an alternative program was noted. Participation in any program of heroin or methadone substitution administered by a public organization was considered participation in an opioid maintenance program.

A patient's behavior was considered to be non-compliant if it was disruptive and violated hospital rules according to the charts, i.e., if the patient did not comply with diagnostic measures or adhere to therapeutic measures, left the hospital against medical advice, continued intravenous drug use during hospitalization, smoked in the room, or assaulted hospital staff.

The study was approved by the Ethics Committee of the Cantons Basel-Stadt and Basel-Land (EKBB).

The outcome parameters were the appropriateness of the prescribed therapy (the choice of treatment and the duration of treatment with respect to standard guidelines for treating specific infectious diseases), the readmission rate (overall rate and the rate due to relapse), in-hospital mortality (both infectious-disease related and non-infectious-disease related), and outpatient mortality (death registries for the Canton Basel-Stadt).

Relapse was defined as infection by the same microorganism(s) responsible for the initial infection at the same site within 30 days after the patient was discharged from the hospital. For osteomyelitis, the time limit for relapse was extended to twelve months.

Briefly, the most important definitions were: skin and soft tissue infections according to the practice guidelines of the Infectious Diseases Society of America 16; pneumonia according to the European guidelines 17; endocarditis according to the Dukes criteria 18; septic thrombosis was defined by positive blood cultures and the detection of a thrombus in imaging 1920, Finally, sepsis was defined by an infection and two or more SIRS (systemic inflammatory response syndrome) criteria 21. Other diagnoses of infectious diseases were according to the Centers for Disease Control and Prevention (CDC) guidelines 22.

Major psychiatric disease, defined as major depression, schizophrenia, or personality disorders, was noted if mentioned in the final medical report as assed by a board certified psychiatrist.

Complications were defined as non-compliant patient behavior, renal failure, septic shock, or admission to the intensive care unit (ICU) during hospitalization.

The appropriateness of the prescribed therapy was evaluated according to written internal guidelines of the infectious diseases division, which are closely adapted from international guidelines (The Sanford Guide To Antimicrobial Therapy 2007, 16172324). Antibiotic therapy for septic thrombosis is based on expert opinions 23. In our study, therapy for septic thrombosis was defined as adequate if a minimum duration of four weeks of treatment was planned and the antibiotic was given intravenously for the first 2 weeks 20 according to the guidelines for treating right heart endocarditis 2425.

The appropriateness of treatment was analyzed by at least two investigators, which were not informed about the outcome of the cases at time of analysis. If there was no agreement, a third investigator was consulted. Whenever a pathogen was isolated, antibiotic susceptibility testing was considered for treatment. Our laboratories are working according to Clinical and Laboratory Standards Institute (CLSI) guidelines.

A multiple logistic regression model was used to describe the dependence of patient compliance and in-hospital mortality on covariates, which were chosen according to clinical concerns. Adjustment for multiple hospitalizations of individual patients was made by using a generalized estimating equation (GEE) approach. Parameters were first estimated by using ordinary logistic regression, ignoring correlations between different hospitalization periods for a single patient (under "working independence"). Standard errors were then adjusted for repeat hospitalizations by using a robust sandwich estimator. The same statistical approach was used for univariate analyses. Results are given as odds ratio (OR) and 95% confidence interval (CI).

During six years, between January 2001 and December 2006, there were 2002 hospitalizations of IVDU (Figure 1). In 420 of these hospitalizations, a specialist in infectious diseases was consulted. Of these 420 cases, 76 (18.1%) were excluded from the study. Demographic data and laboratory results for the 344 hospitalizations of 216 patients are summarized in Table 1. Although the majority of IVDU (93.7%) were enrolled in an opioid maintenance program, 98.4% also used illicit drugs in addition to their maintenance doses. The patients in this study population suffered from diverse co-morbidities and malnutrition as indicated by the rather low body mass index.

The median time to the first consultation by an infectious diseases specialist was 2 days (range 0–92 days). In median, two consultations (range 1–10) were performed per patient.

The distribution of infectious diseases that were diagnosed in this population, the number of infections per hospitalization of a patient, the number of surgical interventions, and the comorbidities that were observed in the patients are summarized in Table 2. The most frequent diagnoses were skin and soft tissue infections followed by pneumonia and endocarditis. A high number of septic thromboses was diagnosed (n = 36, 10.5%). In addition, 23 non-septic thromboses of the deep venous system were diagnosed. Sepsis or severe sepsis due to different infections was documented in 35.5% (n = 122) cases, with no obvious clinical focus in 7.6% (n = 26). Of the 'various infections' listed in Table 1 (n = 75, 21.8%), the most frequently encountered were upper respiratory tract infections (n = 12), infections by Candida species (n = 10), urinary tract infections (n = 10), abdominal infections (n = 6), tuberculosis (n = 5) and sinusitis (n = 4). Eighty-six (25%) of the cases were characterized by two or more simultaneous infections. Viral co-infections, i.e., hepatitis C, HIV, and chronic hepatitis B, were common as comorbidities. Like IVDU in an earlier study 1126, close to 80% of the patients in this study were positive for hepatitis C.

Overall, a pathogen could be isolated in 78.8% of the cases (Table 3). The percentage of microbiologically documented infections was 100% for bacterial endocarditis and septic thrombosis, 96% for bone and joint infections, 85.7% for primary sepsis without focus, 77.7% for skin and soft tissue infections, and 72% for pneumonia. The percentage of microbiologically documented infections for all other types of infections together was 62.3%.

S. aureus was the most frequently isolated pathogen, followed by beta-hemolytic and viridans streptococci (Table 3). S.aureus was the most frequent pathogen isolated in all categories of infections: endocarditis (76% of cases with a primary pathogen identified), bone and joint infections (58%), primary sepsis (57%), septic thrombosis (56%), skin and soft tissue infections (45%), pneumonia (35%) and in other infections (35%). Polymicrobial infections were observed in 18.6% of cases.

A beta-lactam antibiotic, most frequently amoxicilline/clavulanate, was used for treatment in 84.9% of the cases (Table 3). The median duration of intravenous and oral antibiotic therapy was 12 days (range 0–117 days) and 1 day (range 0–36 days), respectively.

Treatment according to standard guidelines was possible in 270 (78.5%) of the 344 hospitalizations. A best alternative therapy was prescribed as recommended by an infectious disease specialist for non-compliant patients in 39 (11.3%) cases. Therapy that was not according to an infectious disease specialist advice was prescribed in 35 (10.2%) cases (Figure 1).

For some infections, e.g., for skin and soft tissue infections, sepsis, or pneumonia, 47.1% of the cases were scheduled for additional outpatient therapy after an adequate or best alternative in-hospital course of treatment (Table 4). In 12.2% of the hospitalizations, patients were discharged against medical advice either with or without therapy.

Complications occurred in 153 hospitalizations (44.5%). Complications noted were acute renal failure (8.4%) and septic shock (3.2%). In 21.5% of the hospitalizations, the patients were admitted to the ICU (median length of ICU stay = 3 days). The presence of pneumonia or endocarditis most often led to ICU admission–36% of all cases of pneumonia and 29.6% of all cases of endocarditis. Non-compliant patient behavior was the most frequent complication (19.8%), and a multiple logistic regression analysis (Table 5) revealed a significant association of non-compliance with psychiatric disorders (OR = 2.4, CI 1.1–5.1, p = 0.03). In hospitalizations of non-compliant patients, the patient was able to follow the prescribed therapeutic program in accordance with standard guidelines in 27.9% of the cases.

The overall 30-day readmission rate was 13.7% (Table 4). Among readmitted cases, 27.7% had a relapse (overall relapse rate 3.8%). The highest relapse rates occurred with septic thrombosis (5.6%) and endocarditis (3.7%). Both non-compliant patient behavior (OR = 3.7, CI 1.3–10.8, p = 0.02) and non-adherence to treatment guidelines (OR = 3.3, CI 1.1–9.7, p = 0.03) were associated with a significant increase in the relapse rate in univariate analysis. Because of the low number of relapses, no multiple (adjusted) logistic models were performed.

Overall in-hospital mortality was 6.4% (n = 22, Table 4). The highest mortality rates were associated with bacterial endocarditis (13%), pneumonia (12%), and sepsis (9.6%). The age of the patient was the only independent risk factors for infection-related, in-hospital mortality (Table 5). State death record information was available for 199 (92.1%) of the 216 patients, which allowed the evaluation of mortality rates after discharge from the hospital (a total of 590 person-years of follow-up). The median observation time was 35.6 months (range 1.6–73.9 months). A total of 49 (24.6%) patients died during the observation period. Of these 49 patients, 27 (55.1%) died after discharge from the hospital, 3 of them within 30 days (2 for whom death was possibly infection related, 1 who died of a trauma). The remaining 24 deaths occurred more than 30 days after the last date of discharge (median 232 days).

There was no significant difference in non-compliant behavior, relapse- or readmission rate between HIV-infected and HIV-negative patients. The association between infection-related deaths and HIV status during hospitalization was non-significant (OR = 1.9, CI 0.8–4.9, p = 0.161). Overall in hospital mortality (OR = 2.4, CI 1.0–5.6, p = 0.06) and mortality in the follow up period (OR = 2.4, CI 1.3–4.4, p = 0.006) was increased for episodes with HIV-infected patients.