Community-Acquired Staphylococcus aureus Bacteremia Among People Who Inject Drugs: A National Cohort Study in England, 2017–2020

Abstract Background People who inject drugs (PWID) are at increased risk of community-acquired Staphylococcus aureus bacteremia (CA-SAB), but little is known about clinical outcomes of CA-SAB in PWID compared with the wider population of patients with CA-SAB. Methods Three national datasets were linked to provide clinical and mortality data on patients hospitalized with CA-SAB in England between 1 January 2017 and 31 December 2020. PWID were identified using the International Classification of Diseases, Tenth Revision code for “mental health and behavioral disorder due to opioid use” (F11). Multivariable logistic regression was used to estimate adjusted odds ratios (aORs) for associations of PWID with 30-day all-cause mortality and 90-day hospital readmission. Results In 10 045 cases of CA-SAB, 1612 (16.0%) were PWID. Overall, 796 (7.9%) patients died within 30 days of CA-SAB admission and 1189 (11.8%) patients were readmitted to hospital within 90 days of CA-SAB. In those without infective endocarditis, there was strong evidence of lower odds of mortality among PWID compared with non-PWID (aOR, 0.47 [95% confidence interval {CI}: .33–.68]; P < .001), whereas there was no association in CA-SAB case fatality with endocarditis (aOR, 1.40 [95% CI: .87–2.25]; P = .163). PWID were less likely to be readmitted within 90 days of CA-SAB (aOR, 0.79 [95% CI: .65–.95]; P = .011). Conclusions In this large cohort study of patients with CA-SAB in England, PWID had lower odds of death in the absence of endocarditis and lower odds of readmission within 90 days compared to non-PWID patients. This study highlights the overrepresentation of PWID among patients with CA-SAB nationally.

The United Kingdom (UK) has the highest rate of drug use in Western Europe, and it is estimated that among 15-to 64-year-olds in England, 87 000 (roughly 0.2%) inject opiates and/or crack cocaine [1].Injection drug use is associated with an increased risk of acquiring a wide range of infections including community-acquired Staphylococcus aureus bacteremia (CA-SAB) [2][3][4][5][6].Staphylococcus aureus may gain entry to the bloodstream through direct inoculation via contaminated needles, through direct injection from an area of skin colonization, or through dissemination from skin infection.The opiate injectate can also be contaminated with S. aureus [7].CA-SAB carries significant mortality of 10%-30% and lengthy hospitalization [8][9][10].It is associated with a range of complications including severe skin and soft tissue infections (SSTIs), osteomyelitis, and infective endocarditis (IE) [11].There are also high rates of recurrent S. aureus infection; the median time from completion of therapy to relapse or reinfection was 36 days (range, 10-190 days) and 99 days (range, 45-194 days), respectively [12,13].Approximately 22% of patients with CA-SAB are readmitted to hospital within 30 days [14].
In the UK and United States (US), a steady increase in cases of SAB linked to injection drug use was observed from 2012 until 2019 [1].Over time, injection drug use leads to thrombosis of superficial veins; therefore, older PWID may be more likely to use sites with heavily colonized skin such as veins in the groin [3,15].Complex psychosocial issues such as social deprivation, homelessness, and psychiatric comorbidity, compounded by stigma associated with injection drug use and its criminalization, may adversely impact access to healthcare and lead to disproportionately poor health outcomes [16,17].
Despite PWID making up a remarkable proportion of patients hospitalized with CA-SAB, few studies have directly compared their clinical outcomes with non-PWID patients or explored factors associated with poor clinical outcomes in this medically and socially vulnerable patient group.Available studies are limited to small single-center studies often with insufficient or no adjustment for potential confounders [16][17][18][19].Data on clinical trajectory would be of value to clinicians and for planning public health measures to prevent CA-SAB in PWID.This study aims to determine if injection drug use is associated with poor clinical outcomes following CA-SAB in England.

Data Sources and Patients
All S. aureus blood culture isolates with sample dates between 1 January 2017 and 31 December 2020 in England were identified through the UK Health Security Agency's Second-Generation Surveillance System (SGSS).SGSS is a voluntary laboratory surveillance system which receives reports from 98% of National Health Service (NHS) laboratories [20].Reporting of all cases of SAB through SGSS was mandatory nationally throughout the study period.These data were linked to public NHS hospitalization data from Hospital Episode Statistics (NHS digital, London, UK) by matching deterministically on patient NHS number, date of birth, and sex [21].Deaths were identified and linked by NHS Digital's Demographic Batch Service from the Office for National Statistics [22].In the UK, a small minority of the population (<10%) accesses private healthcare; however, emergency presentations such as for sepsis are almost exclusively to NHS hospitals, particularly among PWID [23][24][25].

Case Definition and Outcomes
CA-SAB was defined as a blood isolate taken in the period between 14 days prior and 48 hours after hospital admission that was positive for S. aureus.The first admission that met these criteria was considered the index admission for CA-SAB, and only the first admission was included in analyses.Transfers between hospitals within the same inpatient episode were treated as 1 admission.
The International Classification of Diseases, Tenth Revision (ICD-10) code for "mental health and behavioral disorder due to opioid use" (F11) was used as a proxy for PWID, as described previously [26,27].Extracted data items included age, sex, ethnicity, region, blood culture date, methicillin resistance (defined as S. aureus that was either methicillin or cefoxitin resistant), clinical infection syndrome, and medical, social, and psychiatric comorbidities (according to ICD-10 codes, see Supplementary Table 1), admission and discharge dates, and date of death [28].
The primary outcome measure was 30-day all-cause mortality (from the date of index CA-SAB).All-cause readmission within 90 days of the index CA-SAB date was a secondary outcome.Patients who died within 90 days or remained in hospital >90 days were regarded as no readmission in the readmission analysis.

Statistical Methods
All statistical analyses were performed using Stata IC 15.0 (StataCorp, College Station, Texas).Differences in clinical, microbiological, and demographic features comparing PWID with all other patients were tested using the χ 2 test (for categorical variables) and Wilcoxon rank-sum tests (for age and length of stay [LOS]).Odds ratios (ORs) were used to assess the crude association between PWID and each outcome.Multivariable logistic regression models were built to include the following a priori confounders: age, sex, ethnicity, methicillin sensitivity, diabetes, chronic kidney disease (CKD), malignancy, psychiatric illness, homelessness, and alcohol dependence.These factors have previously been shown to have an impact on clinical outcomes (see Supplementary Figure 1 for the conceptual framework) [8,12,13,[29][30][31][32][33].Effect modification by age, sex, homelessness, IE, and SSTI was tested by likelihood ratio test in univariable and multivariable models.Complete case analyses were conducted.

Length of Stay
For the CA-SAB cohort overall, the median LOS was 13 days (IQR, 5-24 days).PWID tended to have a longer duration of hospital stay than non-PWID patients, with a median of 15 days (IQR, 5-27 days) and 13 days (IQR, 5-23 days), respectively (P < .001).

DISCUSSION
People who inject drugs were heavily overrepresented in this large national patient cohort representing 16% of hospitalized patients with CA-SAB compared to 0.2% in the general population [1].CA-SAB is a common infection with associated remarkable morbidity and mortality, but little is known about clinical outcomes in this marginalized patient group.Our data show that PWID had lower odds of death in the absence of endocarditis and were less likely to be readmitted to hospital within 90 days compared to non-PWID patients.This lower mortality was possibly because bacteremia in PWID largely relates to specific behavior rather than underlying illness.The age of PWID in this study was younger than non-PWID patients, consistent with a median age of 41 years (IQR, 36-48 years) for the PWID population in England [6,34].This may account for the lower prevalence of medical comorbidities observed among PWID, as has been reported previously [19].Diabetes, HIV, CKD, and malignancy have been associated with increased risk of recurrent infection and death following SAB [12,13,33].In our cohort overall, only HIV and malignancy were significantly associated with increased unadjusted odds of death, and diabetes and CKD were associated with lower odds of readmission.
PWID may also be more prone to transient bacteremia or bacteremia due to less virulent strains of S. aureus.Direct inoculation of S. aureus into the bloodstream bypasses primary defence mechanisms including the skin, immune cells in the subcutaneous tissues, and the blood vessel wall.This may allow relatively poorly pathogenic strains of S. aureus (lacking in virulence factors) to produce a bacteremia, with milder associated clinical syndromes.This issue has recently been investigated in a case-control study conducted in Missouri, US, by Marks et al [35].The authors found that PWID had lower 1-year mortality associated with SAB (14.3% vs 29.7%).Phylogenetic and comparative genomic analyses demonstrated that intravenous drug use-associated bloodstream infections (IDU-BSIs) grouped into multiple unique clonal clusters and that S. aureus strains lacking in classical virulence factors were overrepresented among IDU-BSIs.Despite the lower mortality observed, they also described considerable complications and morbidity associated with CA-SAB in PWID including prolonged bacteremia and higher prevalence of IE [18].
Only 1 previous study made direct comparisons of clinical outcomes in PWID and non-PWID patients with CA-SAB.This was a single-center study by McClellan et al of 248 patients in Oregon, US [19].The authors reported a weak trend toward lower risk of 90-day mortality (4.3% vs 9.6%, P = .170)and increased risk of readmission within 90 days (46.3% vs 39.8%, P = .154)among PWID compared to non-PWID patients.While PWID were more likely to have MRSA bacteremia in both studies, the prevalence of MRSA in SAB in the US (48.5%) is much higher than in the UK (6.8% in this study) [14,36].Methicillin resistance has been associated with increased LOS, readmission, and death in SAB [13,14].However, while patients with MRSA in our study had increased LOS compared to those with methicillin-sensitive S. aureus, there was no association with mortality, and they had lower odds of readmission.
One study based in London noted that PWID had frequent hospital admissions (up to 12 over a 4-year period) [18].PWID have also been reported to have high rates of discharge from hospital against medical advice-25.7% in PWID versus 1.1% in non-PWID in a US study and 15% in PWID in a UK study [16,19].However, we observed a lower risk of readmission among PWID possibly because LOS was slightly longer for PWID as was reported by McClellan et al [19].In the UK, CA-SAB is generally treated in hospital with intravenous (IV) antibiotics for a minimum of 2 weeks, after which patients requiring longer courses can complete IV therapy via outpatient parenteral antibiotic therapy (OPAT) [29].However, OPAT is often not considered possible for PWID due to challenges in delivering IV therapy [10,16].Clinicians may have reservations about discharging patients who inject drugs with venous access in situ due to concern that the cannula could be used to inject drugs.This, along with the higher prevalence of psychosocial issues experienced by PWID, may contribute to delays in hospital discharge.
A high burden of homelessness was observed within the PWID population (10.2% vs 1.1% in non-PWID), in keeping with findings from the Washington-based study by Marks et al (12% vs 0%) [35].However, even this figure is likely to be an underestimate.A survey of PWID in England in 2008 found that 86% had been homeless at some stage and 59% over the past month [6].PWID may have different experience of homelessness, such as a higher rate of street homelessness, which might differentially affect clinical outcomes in SAB [2][3][4]32].In our overall cohort, homelessness was associated with lower odds of death but was not associated with readmission.This unexpected finding may relate to the younger age of the homeless population (only 12% of homeless people are aged >55 years); however, in 2021 the average age of death of men and women experiencing homelessness was 45 and 43 years, respectively [37].However, this finding may simply reflect the broad definition of homelessness that we used or misclassification due to inaccurate coding.
The prevalence of HIV observed among PWID (0.2%) was unexpectedly lower than among non-PWID (1.2%) and lower than estimates of the prevalence of HIV among PWID in England, which range from 0.8% to 1.1% [1,6].Our study relied on ICD-10 coding for classification of HIV status, and low uptake of screening may have led to an underestimate of the burden of HIV in PWID.In 2021, 18% of PWID who were in contact with services and currently inject reported never having had an HIV test [1].Disruptions to HIV testing during the pandemic may have contributed to poorer screening during the study period.In 2021, immediately after our study period, the UK government's Zero HIV action plan recommended switching to an opt-out HIV testing strategy in areas with the highest HIV prevalence [38].Many acute urban centers now offer opt-out HIV testing, so the HIV prevalence in this study may not be reflective of the current picture.PWID were more likely to have endocarditis, SSTI, and pneumonia, consistent with single-center studies [18,19].For example, McClellan et al reported that, among hospitalized patients with SAB, PWID had higher rates of IE (31.4% vs 8.5%, P < .001)and of SSTI (25.7% vs 14.8%, P = .02)than non-PWID [19].In patients with SAB, endocarditis and pneumonia are associated with increased risk of SAB recurrence, readmission, mortality, and increased LOS [8,9,12,33].The higher prevalence of both conditions observed among PWID may reflect delays in presentation, suggesting outcomes could be improved with better access to healthcare [6,[39][40][41].PWID may also be affected by a different spectrum of SSTIs (eg, groin abscess and chronic fistulae) which may affect clinical outcome [6,15,42].
A considerable strength of our study is that the linked data from 3 national datasets we used offer national coverage inclusive of a broad demographic, enhancing the generalizability of our findings.Use of national hospitalization and mortality datasets enabled identification of readmissions and deaths across different NHS Trusts or geographic regions, which is particularly relevant because PWID may be a more mobile population due to higher risk of homelessness.Potential confounding by comorbidities known to be risk factors for mortality and readmission in SAB was mitigated by adjusting for these factors.
A limitation of our study is that the national datasets lack information on individual patient management, such as antibiotic regimens and infection specialist review.These treatment factors are known to influence outcomes in SAB, and it is possible that they differ between PWID and non-PWID patients [8,9,11,[29][30][31]43]. Clinicians may, for example, opt for oral antibiotics for PWID due to difficulties with IV access [16].Poorer clinical outcomes due to these alternative regimens, however, would act in the opposite direction to our finding of lower 30-day mortality.Higher mortality rates can lead to apparently lower readmission rates (ie, competing risks), but both mortality and readmission rates were lower in PWID in our study.Potential unmeasured confounders include smoking, socioeconomic status, and differences between urban and rural settings.We also did not have access to data on whether endocarditis was left-sided or right-sided.This represents another source of potential confounding because the latter is known to be particularly associated with injection drug use and has better clinical outcomes.
The F11 ICD-10 code was used as a proxy to identify PWID.In England, the use of ICD-10 codes for mental health disorders has been validated with reasonable certainty [44].Additionally, according to the National Statistics in England, only minority (circa 2%) of nonopiate users inject drugs in 2021 [34].When used in conjunction with inclusion criteria of admission due to SAB, this code is likely to be more specific for individuals who are currently injecting.The proportion of CA-SAB cases designated as PWID in our study is comparable to previous reports of 10%-20% in England over this period [1].
The final year of our 4-year study period included the first 9 months of the coronavirus disease 2019 (COVID-19) pandemic in England.Public health measures may have impacted PWID, potentially limiting access to safe injecting equipment and blood-borne virus testing [1].However, the COVID-19 pandemic represented only 9 months of the 48-month study period and is unlikely to have had a significant effect on our overall findings.
Our study highlights the overrepresentation of PWID among patients with CA-SAB and provides information that could be helpful for health services and public health intervention planning.A proportion of CA-SAB is likely to be preventable through public health actions targeted at PWID, including education regarding risks of bacterial infection and safe injecting practices, improved access to opiate substitution programs, and provision of clean needles and wound care packs [1-3, 6, 27, 34].
Prospective observational studies with data on injecting behavior and detailed care pathway and treatment data could inform the design of complex intervention studies aimed at reducing the burden of infection experienced by PWID patients by addressing the myriad of health and psychosocial factors they face.Further studies delineating S. aureus lineages and virulence factors in the PWID population in the UK as well as preclinical studies exploring lineage-dependent pathogenesis would be of value.

SAB
Outcomes in People Who Inject Drugs • CID 2024:78 (15 June) • 1443 Clinical Infectious Diseases M A J O R A R T I C L E

Figure 1 .
Figure 1.Flowchart of data linkage and exclusions for study populations of main and sensitivity analyses.Abbreviations: HES, hospital episode statistics; ONS, Office for National Statistics; PWID, people who inject drugs; SAB, Staphylococcus aureus bacteremia; SGSS, Second-Generation Surveillance System.

Table 1 . Demographics, Methicillin Resistance, Comorbidities and Associated Clinical Infection Syndromes by Opiate Dependence Status in Patients Hospitalized With Community-Acquired Staphylococcus aureus Bacteremia in England, 1 January 2017 to 31 December 2020 (N = 10 045)
Cases with missing values for sex (n = 2) and MRSA (n = 1534) have been excluded.Abbreviations: HIV, human immunodeficiency virus; CKD, chronic kidney disease; MRSA, methicillin-resistant Staphylococcus aureus; PWID, people who inject drugs.a P values derived from χ 2 tests.

Table 3 . Outcomes of People Who Inject Drugs (PWID) With Community-Acquired Staphylococcus aureus Bacteremia (CA-SAB) Compared to Non-PWID Patients With CA-SAB in England (N = 10 045)
Cases with missing values for sex (n = 2) and methicillin-resistant Staphylococcus aureus (n = 1534) have been excluded.Abbreviations: CI, confidence interval; IE, infective endocarditis; OR, odds ratio; PWID, people who inject drugs.aCrude and adjusted ORs estimated using logistic regression; overall and stratum-specific P values derived from the Wald test.Adjusted by age, sex, ethnicity, methicillin sensitivity, diabetes, chronic kidney disease, malignancy, psychiatric illness, homelessness, and alcohol dependence (human immunodeficiency virus excluded due to data sparsity). b