High prevalence of albuminuria amongst people who inject drugs: A cross-sectional study

Albuminuria is a key biomarker for cardiovascular disease and chronic kidney disease. Our study aimed to describe the prevalence of albuminuria amongst people who inject drugs in London and to test any potential associations with demographic characteristics, past diagnoses, and drug preparation and administration practices. We carried out a cross-sectional survey amongst people who use drugs in London. The main outcome measure was any albuminuria including both microalbuminuria and macroalbuminuria. Three-hundred and sixteen samples were tested by local laboratory services. Our study initially employed point-of-care testing methods but this resulted in a high number of false positives. Our findings suggest the prevalence of albuminuria amongst PWID is twice that of the general population at 19% (95%CI 15.3–24.0%). Risk factors associated with albuminuria were HIV (aOR 4.11 [95% CI 1.37–12.38]); followed by overuse of acidifier for dissolving brown heroin prior to injection (aOR 2.10 [95% CI 1.04–4.22]). Albuminuria is high amongst people who inject drugs compared to the general population suggesting the presence of increased cardiovascular and renal pathologies. This is the first study to demonstrate an association with acidifier overuse. Dehydration may be common amongst this population and may affect the diagnostic accuracy of point-of-care testing for albuminuria.

as a person with an low eGFR (<60) without albuminuria 9,10 . Albuminuria has been associated with additional microvascular pathologies (e.g. involving the skin, brain, lungs) which suggests that albuminuria may also be a marker of systemic microvascular disorder 11 . Albuminuria is associated with cardiovascular risk factors including; overweight/obesity, diabetes, high blood pressure, and tobacco smoking 10 .
Population-level estimates reveal a high prevalence of albuminuria in the UK. A 2004 study among those aged 40-79 years in Norfolk, UK estimated the prevalence of microalbuminuria as 11.2% (13.9% in women, 8.1% in men) and macroalbuminuria as 0.7% (0.7% in women, 0.8% in men) 12 . The 2009 and 2010 Health Surveys for England estimated that 8% of adults surveyed had abnormal levels of albumin in their urine (7.9% microalbuminuria, 0.5% macroalbuminuria) 10 .
This paper aims to describe the prevalence of albuminuria among PWID, and to inform the clinical use and interpretation of albuminuria testing amongst this population.

Methods
We undertook a cross-sectional survey, as part of the UK National Institute for Health Research-funded Care & Prevent study, aimed at assessing the evidence for AA amyloidosis amongst PWID 13 . We developed a computer-assisted questionnaire to identify patterns of drug use and potential risk factors for SSTIs and AA amyloidosis. The questionnaire was conducted with current or past PWID aged 18 years and over who were recruited in London at eight drug treatment services and a mobile health service working with homeless people (UCLH Find & Treat Service). Participants completed the questionnaire followed by a urine screen for albuminuria. Participants were asked questions relating to: demographics, injecting history, injection practices, HIV and hepatitis C virus (HCV) status, SSTI history, and other conditions associated with albuminuria. Potential confounders were age, sex, and tobacco smoking.
Urine was initially tested using laboratory urinalysis and/or point of care (POC) testing using CLINITEX Microalbumin Reagent Strip (Siemens Healthcare GmbH), however, the POC testing yielded a high number of false positives (See Text Box 1) and was abandoned in favour of laboratory urinalysis. Albumin levels between 2.8 and 29.9 mg/mmol were considered abnormal (i.e. microalbuminuria), and ≥30 mg/mmol were deemed highly abnormal (i.e. macroalbuminuria). Participants with macroalbuminuria were referred to the University College London (UCL) National Amyloidosis Centre at Royal Free Hospital in North London for diagnostic assessment for AA amyloidosis.
Text Box 1. We initially used CLINITEK Microalbumin 2 Strips in a CLINITEK Status+ Analyser to determine the albumin-to-creatine ratio (ACR) of urine samples. CLINITEK Microalbumin 2 Strips were selected as they are specified for use with spontaneous urine samples and measure both albumin and creatinine values, enhancing the specificity of results. Despite manufacturer guidelines and reviews assuring adequate specificity and sensitivity for POC testing, we found a discrepancy in 20 of the 45 samples that were also sent for laboratory testing. While POC results indicated albuminuria in 58% of cases (26 micro and 1 macro), laboratory results for the same samples showed 13% with albuminuria (6 micro and 1 macro). (See Figure 1) Given this discrepancy we abandoned POC testing. We brought on an additional site with a laboratory pathway, and developed a collaboration with the UCLH Find & Treat Service outreach team, enabling us to recruit through homeless hostels via their outreach team and to use their private laboratory pathway.
The discrepancy between POC samples and laboratory samples is likely due to the highly concentrated urines of many participants. Concentrated urine has a very high creatinine level, which would have exceeded the microalbumin strip cut-off point -thus creating a false positive. Highly concentrated urines were also observed on handling, indicating inadequate water consumption and possible dehydration in this population. Aside from the health implications of dehydration, our findings indicate that semi-quantitative POC testing should be avoided or used with extreme caution with this population. Clinitek results for ACR Laboratory results for ACR All participants were provided with study information sheets and gave written consent prior to answering the questionnaire and providing a sample for urinalysis. Participants were reimbursed for their time with a £10 gift voucher. All methods were carried out in accordance with the relevant guidelines and regulations. This study has been reported against the STROBE reporting checklist for observational studies 14 . patient and public involvement. The survey was developed in collaboration with the Lambeth and Camden Drug User Forums. Members of the National Amyloidosis Centre Patient Network helped develop the participant information sheets. Groundswell, a local homelessness charity 15 , provided input to the study as well as peer support 16,17 (See GRIPP2 Reporting Checklist Appendix 1).
Sample size. We aimed to collect urine samples from 400 participants based on feasibility (rates of service attendance) and an estimated 5% prevalence of proteinuria (95% CI, 3-7%).
Quantitative variables. The main outcomes of interest were any albuminuria; microalbuminuria (a biomarker for other health outcomes, such as cardiovascular disease) and macroalbuminuria (a biomarker for AA-amyloidosis). Exposures were chosen a priori as plausible factors in the causal chain and were based on the literature, our clinical experience, and our prior research involving PWID. Exposures include: injecting practices, location of injection sites, number of injection sites used, SSTI history, severity of past or current SSTI, housing status, overuse of acidifier (for dissolving drugs prior to injection), and self-reported lifetime diagnoses of health conditions (e.g. necrotising fasciitis, COPD, sepsis, endocarditis). Smoking status was added to the survey following commencement of data collection owing to the unexpectedly high prevalence of microalbuminuria, and is thus only available for 75% of the total sample, and 95% of the subsample of participants for whom we have laboratory results.

Statistical analysis.
Associations between potential risk factors and albuminuria were tested using crude logistic regressions. Bivariate associations found to have a significance of ≤0.1 were introduced into the multivariate logistic model to examine independent associations with albuminuria. We used a 95% confidence interval for the calculation of population point estimates of albuminuria. We further estimated a ratio of the prevalence of albuminuria in our PWID sample to the prevalence in the general population using data from the Health Survey for England 2016, which included urinalysis. We fit a poisson regression model on the combined data with albuminuria as the dependent variable and drug injection as the independent variable, adjusting for age group and sex. We then additionally adjusted for smoking status (current, former, or non-smoker).

Results
Of the 455 participants who completed the survey, 442 consented to urinalysis. Three-hundred and sixteen laboratory urinalyses were returned. The 139 samples not included in the analysis were either tested using the POC CLINITEX Microalbumin Reagent Strips only (e.g. for services with no laboratory pathway) or they were rejected www.nature.com/scientificreports www.nature.com/scientificreports/ by the laboratory (e.g. the sample container was damaged). We compared the personal characteristics of those from whom we obtained a sample to those from whom we did not using the Mann-Whitney U test (for age) and Pearson χ-squared and found no significant difference between the two groups (See Appendix 2).
Main results. In a series of crude analyses the following associations were found with albuminuria (See Table 4  Based on these associations we performed logistic regression using listwise deletion of observations with missing smoking status and adjusted for a priori confounders (i.e. age, gender, and current smoking). We found higher odds of having albuminuria amongst those who: reported using more than half a sachet of acidifier per £10 of heroin ( In order to determine the age-adjusted prevalence ratio we compared our sample to the Health Survey for England data for 2016. Accounting for smoking status and stratified by age, we excluded the youngest (<25) and oldest (65+) age groups as they included small numbers of observations. The adjusted prevalence ratio suggests that albuminuria is twice as prevalent (  www.nature.com/scientificreports www.nature.com/scientificreports/ Discussion Statement of principal findings. This study sought to determine the prevalence of albuminuria amongst PWID in London. Previous studies have estimated that albuminuria in the general adult population in the UK is less than 10% 12,18,19 . Our findings suggest prevalence amongst PWID is twice that at 19% (15.3-24.0%). However, the age distribution in our sample differs markedly from that of the general adult population 18 . Adjusting for current smoking the prevalence ratio decreased to 1.65 (95% CI 1.17-2.31), suggesting that about one-third of the increased prevalence relates to smoking. Within our sample of PWID, overuse of acidifier and HIV were important predictors of albuminuria.
HIV is known to be associated with proteinuric renal lesions 20 . Additionally, we hypothesise that persistent overuse of acidifier in injection solution is precipitating venous sclerosis (owing to endothelial stress), thus complicating venous access and resulting in multiple venous injection attempts (and possible accidental subcutaneous injecting) and/or transition to intentional subcutaneous injecting 21 . Subcutaneous injecting introduces substances and/or bacteria directly into tissues which, in turn, causes persistent localised inflammation 1 . Alternatively, we cannot rule out the possibility that over-use of acidifier is directly causing local or systemic inflammation 6 . Chronic inflammation is strongly associated with cardiovascular and renal disease of which albuminuria is a biomarker 9 .
These data suggest that PWID are a high-risk group to develop cardiovascular and renal complications. A recent study of mortality among people who use heroin in South London found 2.8 times the risk of cardiovascular mortality compared to the general population 22 . If the associations (as observed in the general UK population) between albuminuria and future risk of cardiovascular disease are also true for PWID, one would expect the stroke risk in this population to be increased by 90% relative to a person with no albuminuria 23 , whilst the absolute risk of cardiovascular mortality increases by about 30% for a doubling of albuminuria 24 . Strengths and weaknesses of the study. Initially, urine sample testing was carried out using a POC testing machine; however, false positives were identified following confirmatory laboratory testing We believe this to be a result of high levels of dehydration amongst PWID. Our prevalence estimate may have been improved by testing twice (as per NICE guidelines); however, this would have been difficult given our study population 9 . The Health Survey for England albuminuria data were collected based on the same protocol used for the present study (i.e. a single sample) therefore increasing its comparability with our study population 18 .
The absence of a significant association between SSTIs and albuminuria was surprising as we anticipated SSTIs would be a key factor in activating the inflammatory cascade. It is possible that our participants underreported past SSTIs or that our sample was underpowered to draw out this association. Our study may also have been underpowered to detect other meaningful predictor variables. Fnally, we cannot rule out the possibility that HIV and excessive use of acidifier may be contributing to a systemic microvascular disorder.

Unanswered questions and future research.
Future studies should investigate long-term outcomes of PWID realted to cardiovascular and renal risk, though we appreciate the challenges of following up a difficult to capture population. Future trials should investigate whether cardiovascular and renal disease may be delayed by blood pressure lowering therapy or a poly-pill (i.e. a pill containing a combination of several medications) in