Risk factors associated with acute kidney injury in a cohort of 20,575 arthroplasty patients

Full Te ms & Condi ions o access and use can be ound a h p://www. and online.com/ac ion/jou nalIn o ma ion?jou nalCode=io 20 Download by: [Tampe e Uni e si y] Da e: 17 July 2017, A : 23:47 Ac a O hopaedica ISSN: 1745-3674 (P in ) 1745-3682 (Online) Jou nal homepage: h p://www. and online.com/loi/io 20 Risk ac o s associa ed wi h acu e kidney inju y in a coho o 20,575 a h oplas y pa ien s Py y Jämsä, Esa Jämsen, Leo-Pekka Lyy ikäinen, Ja kko Kallio alkama, An i Eskelinen & Niku Oksala To ci e his a icle: Py y Jämsä, Esa Jämsen, Leo-Pekka Lyy ikäinen, Ja kko Kallio alkama, An i Eskelinen & Niku Oksala (2017) Risk ac o s associa ed wi h acu e kidney inju y in a coho o 20,575 a h oplas y pa ien s, Ac a O hopaedica, 88:4, 370-376, DOI: 10.1080/17453674.2017.1301743 To link o his a icle: h p://dx.doi.o g/10.1080/17453674.2017.1301743 © 2017 The Au ho (s). Published by Taylo & F ancis on behal o he No dic O hopedic Fede a ion. View supplemen a y ma e ial Published online: 15 Ma 2017. Submi you a icle o his jou nal A icle iews: 392 View ela ed a icles View C ossma k da a 370 Ac a O hopaedica 2017; 88 (4): 370–376 Risk ac o s associa ed wi h acu e kidney inju y in a coho o 20,575 a h oplas y pa ien s Py y JÄMSÄ 1, Esa JÄMSEN 1,2, Leo-Pekka LYYTIKÄINEN 2,3,4, Ja kko KALLIOVALKAMA 1,2, An i ESKELINEN 1*, and Niku OKSALA 2,5* 1 Coxa Hospi al o Join Replacemen ; 2 School o Medicine, Uni e si y o Tampe e; 3 Depa men o Clinical Chemis y, Uni e si y o Tampe e; 4 Fimlab Labo a o ies; 5 Depa men o Su ge y, Facul y o Medicine and li e sciences, Tampe e Uni e si y Hospi al, Tampe e, Finland. * Sha ed senio au ho ship. Co espondence: Py y.jamsa@ i mne . i Submi ed 2016-10-0. Accep ed 2017-01-23. © 2017 The Au ho (s). Published by Taylo & F ancis on behal o he No dic O hopedic Fede a ion. This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion-Non-Comme cial License (h ps://c ea i ecommons.o g/licenses/by-nc/3.0) DOI 10.1080/17453674.2017.1301743 Backg ound and pu pose — Pa ien s de eloping pos ope a i e acu e kidney inju y (AKI) a e a isk o highe mo bidi y and mo ali y. In a h oplas y pa ien s, many p e- and pe iope a i e ac o s a e associa ed wi h AKI bu some o he isk ac o s a e unclea . We epo he incidence o pos ope a i e AKI, he condi- ions associa ed wi h i , and su i al a es in AKI pa ien s. Pa ien s and me hods — We ob ained da a om 20,575 con- secu i e hip o knee a h oplas ies. Pos ope a i e AKI, occu - ing wi hin 7 days a e he ope a ion, was de i ned using he isk, inju y, ailu e, loss, and end-s age (RIFLE) c i e ia. We analyzed independen isk ac o s o AKI using bina y logis ic eg ession. In addi ion, we e iewed he eco ds o AKI pa ien s and pe - o med a su i al analysis. Resul s — The AKI incidence was 3.3 pe 1,000 ope a ions. We ound p eope a i e es ima ed glome ula i l a ion a e, ASA classi i ca ion, body mass index, and du a ion o ope a ion o be independen isk ac o s o AKI. In ec ions, pa aly ic ileus, and ca diac causes we e he p edominan unde lying condi ions, whe eas hal o all AKI cases occu ed wi hou any clea unde ly- ing condi ion. Su i al a es we e lowe in AKI pa ien s. In e p e a ion — Suppo ing ea lie esul s, exis ing enal insu i ciency and pa ien - ela ed cha ac e is ics we e ound o be associa ed wi h an inc eased isk o pos ope a i e AKI. Fu he - mo e, du a ion o ope a ion was iden i i ed as an independen isk ac o . We sugges ca e ul enal moni o ing pos ope a i ely o pa ien s wi h hese isk ac o s. ■ Acu e kidney inju y (AKI) a ec s 0.5–5.2% o join eplace- men ecipien s and i is an independen isk ac o o ch onic enal impai men , inc eased mo bidi y, and dea h (Ympa e al. 2005, Ja a i e al. 2010, Coca e al. 2012, Kimmel e al. 2014, Pe egaa d e al. 2016). AKI inc eases in-hospi al mo ali y bu he ad e se e ec s o AKI can also occu la e (La ance and Mille 2010). The p eope a i e isk ac o s associa ed wi h pos ope a i e AKI in a h oplas y pa ien s include ele a ed body mass index (BMI), diabe es melli us, ch onic obs uc i e pulmona y dis- ease, li e disease, conges i e hea ailu e, hype ension, and ascula diseases (Ja a i e al. 2010, Weinga en e al. 2012, Bell e al. 2015). Also, age, p eope a i e kidney dys unc ion, and he p eope a i e use o enin-angio ensin axis-blocking medica ion a e associa ed wi h pos ope a i e AKI in a h o- plas y pa ien s (A eline e al. 2009, Kimmel e al. 2014, Nielson e al. 2014, Bell e al. 2015). Al hough he Ame ican Socie y o Anes hesiologis s’ (ASA) physical s a us classi i ca- ion sys em has a connec ion wi h AKI in o hopedic pa ien s (Bell e al. 2015), i is no associa ed wi h AKI in a h oplas y pa ien s (Ja a i e al. 2010, Kimmel e al. 2014). Pe iope a i e ac o s associa ed wi h AKI include gene al anes hesia and blood ans usions (Weinga en e al. 2012). Also, du a ion o ope a ion has been epo ed o be longe in AKI pa ien s, bu no s a is ically signi i can ly so (Ja a i e al. 2010, Weinga en e al. 2012, and Kimmel e al. 2014). The e ec s o non-s e oidal an i-in l amma o y d ugs and concomi- an use o diu e ics, angio ensin-con e ing enzyme inhibi- o s, and angio ensin ecep o blocke s in a h oplas y popula- ions is unclea (Lee e al. 2007, Fou nie e al. 2014). Amino- glycosides (such as gen amycin) ha a e used as in a enous an ibio ic p ophylaxis o in cemen may igge AKI because o hei neph o oxici y (Cu is e al. 2005, Pa ick e al. 2006, Lau and Kuma 2013, Ross e al. 2013, Bell e al. 2014, C ax- o d e al. 2014, Johansson e al. 2016) bu o he wise, he e a e no da a on he condi ions o unde lying easons ha igge pos ope a i e AKI in a h oplas y pa ien s. We assessed he incidence and ac o s associa ed wi h AKI ollowing hip and knee a h oplas y in a la ge Scandina ian coho and epo he speci i c condi ions associa ed wi h AKI. We hypo hesized ha : (1) he Scandina ian popula ion would ha e simila a es o AKI o hose in o he popula- ions; and (2) he ASA classi i ca ion sys em (as a measu e 10927 Ja msa D.indd 37010927 Ja msa D.indd 370 6/20/2017 5:25:14 PM6/20/2017 5:25:14 PM Ac a O hopaedica 2017; 88 (4): 370–376 371 o como bidi y) and du a ion o ope a ion would show an associa ion wi h AKI. Pa ien s and me hods The s udy was pe o med in a la ge publicly unded o hope- dic hospi al specialized in join eplacemen su ge y, wi h an annual numbe o a h oplas ies exceeding 3,000. The s udy popula ion comp ised pa ien s wi h hip o knee a h oplas- ies pe o med a he hospi al be ween Sep embe 2002 and Decembe 2011 (n = 20,575). 2,000 pa ien s we e excluded om he s udy (Figu e 1). The emaining 18,575 pa ien s we e used o he analyses. Demog aphic da a and pa ien in o ma- ion we e ob ained om a p ospec i e join eplacemen da a- base and pa ien adminis a ion da abase. The ollowing da a we e collec ed o analysis: sex, age, indica ion o ope a ion, BMI, ASA classi i ca ion, anes hesia modali y, p ophylac ic an ibio ic, ope a ed join (hip o knee), du a ion o ope a ion, i xa ion me hod (cemen ed, cemen less, hyb id), la e ali y (unila e al o bila e al ope a ion), ype o ope a ion (p ima y o e ision), and use o an ibio ic-imp egna ed bone cemen (Tables 1 and 2). P e- and pos ope a i e se um c ea inine (SC ) le els we e ob ained om he da abase o a local labo a o y ha p o ides ou hospi al, an adjacen uni e si y hospi al, and he majo - i y o he communi ies in he ca chmen a ea wi h labo a o y se ices. The SC le el is ou inely ob ained as pa o he p e-anes hesia e alua ion ca ied ou 1–2 mon hs be o e he ope a ion, bu measu emen s aken wi hin 6 mon hs be o e he ope a ion we e app o ed. I mul iple p eope a i e SC mea- su emen s we e eco ded, he mos ecen SC measu emen was used. Pos ope a i ely, SC was measu ed o clinical indi- ca ions only and no ou inely. In ou s udy, we ook accoun o all pos ope a i e SC measu emen s aken ≤ 7 days a e he ope a ion, which was done in 5,609 ope a ions (30%). These pa ien s had lowe eGFR p eope a i ely (76 mL/min/1.73 m2 s 87 mL/min/1.73 m2), olde mean age (76 yea s s 67 yea s), highe ASA classi i ca ion (median 3 s. 2), and a sligh ly longe mean du a ion o ope a ion (105 min s. 100 min) han pa ien s wi h no SC measu emen done du ing he i s 7 pos - ope a i e days. O hese pa ien s, 39% (2,210) we e male and 22% (1,222) had e ision a h oplas y. The e we e also 5,361 pa ien s (29%) who lacked pos ope a i e labo a o y ollow- up a e discha ge om ou uni because hei home coun y used a di e en labo a o y. We included hese pa ien s in ou analysis o maximize he numbe o AKI cases and he e o e o maximize s a is ical powe . As he cha ac e is ics o hese pa ien s di e ed sligh ly om hose o he pa ien s who we e examined in ou labo a o y (da a no shown), we excluded hese pa ien s om he sensi i i y analysis o elimina e a pos- sible sou ce o bias. We used SC o classi y all he pa ien s in o one o he RIFLE classi i ca ions ( isk, inju y, ailu e) o in o a non-AKI g oup (Bellomo e al. 2007). We assumed ha he pa ien s who we e no es ed o pos ope a i e SC would no ha e had pos ope a i e AKI. To main ain high speci i ci y, hose pa ien s who we e in he isk o AKI class we e classi i ed as no ha ing AKI. We used p eope a i e SC o calcula e he es ima ed glo- me ula i l a ion a e (eGFR) using he CKD-EPI o mula (Le ey e al. 2009). In pa ien s who de eloped AKI (class I o F acco ding o he RIFLE c i e ia), we e iewed he medical eco ds in o de o de i ne he possible p e- and pos ope a i e isk ac o s associ- a ed wi h AKI (Table 4), and he ou come o AKI. We clas- si i ed he ou come as spon aneous e u n o kidney unc ion, loss o kidney unc ion, o end-s age kidney disease. S a is ics Fo he s a is ical analyses, we iden i i ed he AKI cases using he c i e ia desc ibed abo e. Fu he mo e, we used all he non- AKI pa ien s as a con ol g oup. 95% con i dence in e al (CI) o he incidence a e o pos ope a i e AKI was calcula ed using he Wilson sco e in e al. The ela ionship be ween po en ial isk ac o s and AKI was analyzed using uni a i- able bina y logis ic eg ession. Mul i a iable bina y logis ic eg ession analysis was pe o med using he en e me hod o minimize bias. To c ea e a mul i a iable model, we used a di ec ed acyclic g aph (DAG) o es ablish a causal ela ionship be ween a iables and o i nd a minimal adjus men se o min- imize bias. Due o he complexi y o he causal ela ionships among all a iables associa ed wi h AKI, we used he Dagi y ool (Tex o e al. 2011) o c ea e he mul i a iable model. We chose du a ion o ope a ion as an exposu e a iable and AKI as an ou come a iable. The Dagi y model showed ha BMI, i xa ion echnique, bila e al ope a ion, and ope a ion ype and join was he minimal su i cien adjus men se o minimize bias. We also included he ASA classi i ca ions and p eope a- i e eGFR in he mul i a iable model acco ding o clinical expe ience (see Supplemen a y da a) and because hese a i- ables we e in e es ing o ou s udy hypo hesis. We pe o med a sensi i i y analysis ha included only he minimal adjus - men se o make su e ha he esul s emained unchanged when he ASA classi i ca ions and eGFR we e added o he model. A Kaplan-Meie analysis was pe o med o de e mine he e ec o AKI on su i al a es. The esul s we e consid- Analyzed (n = 18,575): – p ima y a h oplas ies, 15,943 – e ision a h oplas ies, 2,632 Eligible hip and knee a h oplas ies Sep embe 2002 – Decembe 2011 (n = 20,575) Excluded (n = 2,000): – lacking p eope a i e SC , 1,031 – eme gency ope a ions, 969 Figu e 1. Exclusion o pa ien s. 10927 Ja msa D.indd 37110927 Ja msa D.indd 371 6/20/2017 5:25:14 PM6/20/2017 5:25:14 PM 372 Ac a O hopaedica 2017; 88 (4): 370–376 e ed s a is ically signi i can when he p- alue was < 0.05. As he incidence o AKI is e y low, all he odds a ios p o ided can be in e p e ed as ela i e isk (RR) unless o he wise s a ed. We used SPSS 21 so wa e o he s a is ical analysis. E hics In Finland, e hical commi ee app o al is no equi ed in e o- spec i e s udies wi h no human subjec s, such as his one. The s udy was accep ed by Pi kanmaa hospi al dis ic (ETL-code 13501) on Ma 1, 2013. Funding and po en ial con l ic o in e es We a e g a e ul o he i nancial suppo o he p ojec gi en by he Finnish A h oplas y Associa ion (Suomen A oplas- iayhdis ys) in 2012. No compe ing in e es s decla ed. Resul s 58 cases o AKI we e iden i i ed in 18,575 pa ien s. Among he 13,214 pa ien s whose specimens we e es ed by ou hos- pi al labo a o y, 44 AKI cases we e iden i i ed and he inci- dence o AKI was 3.3 pe 1,000 ope a ions (95% CI: 2.5– 4.5). O he 58 pa ien s wi h AKI, 43 had inju y- and 15 had ailu e-s age AKI. In uni a iable analysis, he isk ac o s o AKI p eope a i ely we e age, BMI, ASA classi i ca ion, SC , eGFR, and hemoglobin alue (Table 1), and pe iope a i ely he isk ac o s we e ope a ion ype and in a enous an ibi- o ic p ophylaxis (Table 2). The mul i a iable model showed ha du a ion o ope a ion, ASA classi i ca ion, BMI, and p e- ope a i e eGFR we e independen isk ac o s o pos ope a- i e AKI (Table 3). Table 1. Associa ion be ween p eope a i e ac o s and AKI; uni a iable eg ession esul s Incidence 1,000 pe Desc ip o All pa ien s a AKI non-AKI ope a ions OR (95% CI) p- alue Age, median ( ange) 69 (14–102) 76 (41–88) 69 (14–102) 1.05 (1.02–1.08) 0.001 Sex Female 11,650 (63) 33 11,617 2.8 Re e ence Male 6,925 (37) 25 6,900 3.6 1.3 (0.76–2.1) 0.4 Body mass index (missing, n = 1,719) median, ( ange) 28.6 (14–59) 31.6 (24–43) 28.6 (14–59) 1.1 (1.0–1.1) 0.001 < 25 3,551 (21) 6 3,545 1.7 Re e ence 25–30 6,760 (40) 14 6,746 2.1 1.2 (0.47–3.2) 0.7 30–35 4,496 (27) 14 4,482 3.1 1.9 (0.71–4.8) 0.2 > 35 2,049 (12) 12 2,037 5.9 3.5 (1.3–9.3) 0.01 Join Hip 8,821 (47) 23 8,798 2.6 Re e ence Knee 9,754 (53) 35 9,719 3.6 1.4 (0.8–2.3) 0.2 ASA classi i ca ion (missing, n = 132) 1 1,477 (8) 0 1,477 0.0 2 8,427 (46) 9 8,418 1.1 Re e ence b < 0.001 3 8,046 (44) 40 8,006 5.0 5.5 (2.7–11) < 0.001 4 492 (3) 7 485 14.4 16 (5.9–43) < 0.001 P eope a i e SC 70 (25–1,125) 78 (49–150) 70 (25–1,125) 1.004 (1.00–1.01) 0.007 P eope a i e eGFR, median ( ange) c 85 (3–160) 68 (30–108) 85 (3–160) 0.97 (0.95–0.98) < 0.001 1 (> 90 mL/min) 6,519 (35) 9 6,510 1.4 Re e ence 2 (60–89 mL/min) 9,917 (53) 29 9,888 2.9 2.1 (1.0–4.5) 0.05 3 (30–59 mL/min) 2,023 (11) 19 2,004 9.4 6.9 (3.1–15) < 0.001 4 (15–29 mL/min) 81 (0.4) 1 80 12.5 9.0 (1.1–72) 0.04 5 (< 15 mL/min) 35 (0.2) 0 35 0 – – P eope a i e hemoglobin, (missing, n = 72) median ( ange) 138 (76–189) 135 (103–163) 138 (80–189) 0.97 (0.95–0.99) 0.001 P eope a i e anemia d No 16,423 (88) 47 16,376 2.9 Re e ence Yes 2,080 (11) 11 2,069 5.3 1.9 (0.96–3.6) 0.07 Diagnosis P ima y os eoa h i is 15,467 (83) 47 15,420 3.0 Re e ence O he diagnosis 3,073 (17) 10 3,063 3.3 1.1 (0.54–2.1) 0.8 a In his column, numbe s mean numbe o pa ien s and pe cen age in pa en heses unless o he wise s a ed. b ASA classes 1 and 2 we e combined o he eg ession analysis. c eGFR calcula ed using CKD-EPI o mula. d Anemia was de i ned as hemoglobin < 117 g/L in women and < 134 g/L in men. 10927 Ja msa D.indd 37210927 Ja msa D.indd 372 6/20/2017 5:25:15 PM6/20/2017 5:25:15 PM Ac a O hopaedica 2017; 88 (4): 370–376 373 Sensi i i y analysis was pe o med o de e mine whe he hese esul s we e also alid in he p ima y a h oplas y g oup (n = 15,943). Mul i a iable analysis demons a ed he same independen isk ac o s o AKI as iden i i ed in he o iginal analysis. We also epea ed he analysis wi h adjus men o only he a iables in he minimal su i cien adjus men se , and he esul s emained unchanged, con i ming ha bo h du a ion o ope a ion and BMI we e independen isk ac o s o AKI. In he hi d sensi i i y analysis, we excluded all 5,361 pa ien s (14 o 58 AKI cases) wi h a di e en labo a o y egis e in hei communi y and hus possibly lacking a 7-day pos op- e a i e labo a o y ollow-up. This analysis showed he same independen isk ac o s o AKI. A po en ial cause o AKI was iden i i ed in 28 cases. The mos common condi ions we e pos ope a i e in ec ions in 15 pa ien s, ca diac causes in 5 pa ien s, and pseudo-obs uc ion in 5 pa ien s. In 28 cases, no speci i c cause could be iden i i ed e ospec i ely. Mos pa ien s had se e al ac o s ha possibly con ibu ed o he de elopmen o AKI (Table 4). 2 o he AKI pa ien s had no pa ien eco ds conce ning he pos ope a i e pe iod, so he eason o AKI could no be ound. 2 pa ien s wi h AKI unde wen pos ope a i e dialysis o less han 4 weeks and bo h eco e ed hei kidney unc ion. Thus, nei he o hese pa ien s was classi i ed in he loss o kidney unc ion g oup o he end-s age kidney disease g oup. Mo ali y was subs an ially highe in AKI pa ien s han in non-AKI pa ien s h oughou he ollow-up (Figu e 2). Discussion Despi e he ac ha p eope a i e kidney dys unc ion acco d- ing o eGFR is common in ou Scandina ian popula ion (65% o he pa ien s had eGFR < 90 mL/min), AKI is a e (wi h an incidence o 3.3 pe 1,000 a h oplas ies). We ound ha du a- ion o ope a ion, p eope a i e eGFR, ASA class, and BMI we e all independen isk ac o s o pos ope a i e AKI. How- e e , hal o all cases o AKI occu ed wi hou any unde lying Table 2. Associa ion be ween pe iope a i e ac o s and AKI; uni a iable eg ession esul s incidence pe 1,000 Desc ip o All pa ien s a AKI Non-AKI ope a ions OR (95% CI) p- alue Ope a ion ype P ima y 15,943 (85) 44 15,899 2.8 Re e ence Re ision 2,632 (15) 14 2,618 5.3 1.9 (1.1–3.5) 0.03 Knee ope a ion ype Unicondyla 607 (6) 1 606 1.7 Re e ence To al 9,147 (94) 34 9,113 3.7 2.3 (0.31–17) 0.4 Bila e al ope a ion No 16,908 (91) 55 16,853 3.3 Re Yes 1,667 (9) 3 1,664 1.8 0.55 (0.17–1.8) 0.3 P o hesis i xa ion me hod (missing n = 1,410) Cemen less 4,107 (24) 6 4,101 1.4 Re e ence 0.2 Hyb id 2,595 (15) 7 2,591 2.7 1.9 (0.62–5.5) 0.3 To al cemen 10,460 (61) 35 10,425 3.4 2.3 (0.96–5.5) 0.06 Use o an ibio ic-imp egna ed bone cemen (missing, n = 589) Gen amycin 11,764 (94) 40 11,724 3.4 Tob amycin 541 (4) 0 541 0 O he 164 (1) 0 164 0 An ibio ic p ophylaxis (missing, n = 592) Ce u oxime 17,714 (99) 48 17,666 2.7 Re e ence 0.005 Clindamicin 207 (1) 3 204 14.7 5.4 (1.7–17) 0.005 O he 62 (0.3) 1 61 16.4 6.0 (0.82–44) 0.08 Anes hesia modali y (missing, n = 105) Spinal 2,077 (11) 4 2,073 1.9 Re e ence 0.8 Con inuous spinal 1,710 (9) 7 1,703 4.1 6.3 (0.66–61) 0.2 Combined spinal epidu al 14,302 (77) 44 14,258 3.1 3.0 (0.41–23) 0.4 Gene al 287 (2.0) 1 286 3.5 8.3 (0.52–133) 0.6 O he 94 (0.5) 0 94 0 1 Du a ion o ope a ion b (missing, n = 110) median ( ange) 10 (10–68) 12 (6–25) 10 (10–68) 1.0 (1.0–1.1) 0.07 a In his column, numbe s mean numbe o pa ien s and pe cen age in pa en heses unless o he wise s a ed. b Du a ion in 10-min in e als ( ime om incision o end o wound closu e). 10927 Ja msa D.indd 37310927 Ja msa D.indd 373 6/20/2017 5:25:15 PM6/20/2017 5:25:15 PM 374 Ac a O hopaedica 2017; 88 (4): 370–376 cause iden i i ed. The su i al o pa ien s wi h AKI was wo se han ha o non-AKI pa ien s. The incidence o AKI was lowe han ha epo ed by o he g oups (3.3 pe 1,000 s. 5.5 pe 1,000 s. 52 pe 1,000) (Ja a i e al. 2010, Kimmel e al. 2014). This migh be due o lowe BMI (29 s. 32 and 31), a smalle p opo ion o gene al anes hesia (1.5% s. 6.1% and 60%), a la ge p opo ion o ASA class 1 pa ien s (8.0% s. 4.8% and 2.8%), and sho e du a ion o ope a ion (100 min s 157 min s 119 min) in ou coho compa ed o he o he coho s (Ja a i e al. 2010, Kimmel e al. 2014). The incidence o AKI was lowe com- pa ed o Pe egaa d e al. 2016 (3.3 pe 1,000 s. 21.9 pe 1,000). Unlike in ou s udy, hey included mild-s age AKI in hei analysis, which p obably explains he di e ence. We ound ha BMI was an independen isk ac o o AKI, a i nding suppo ed by p e ious s udies (Ja a i e al. 2010. Kimmel e al. 2014). An associa ion has also been shown be ween lowe eGFR and AKI (Ja a i e al. 2010, Kimmel e al. 2014, Bell e al. 2015). In uni a iable analysis, p eope a- i e enal impai men inc eased doubled he isk o AKI by 2- old, when eGFR was lowe han 90 mL/min and inc eased i by 7- old when eGFR was lowe han 60mL/min. The indi- dence o AKI was highe ollowing e ision han a e p ima y su ge y, bu he associa ion was los in mul i a iable analysis, p obably because e ision a h oplas y pa ien s had mo e ac- o s associa ed wi h AKI. In ou popula ion, pa ien s unde go- ing e ision a h oplas y gene ally had a longe du a ion o ope a ion and a highe ASA class. Some case epo s ha e sugges ed ha ancomycin-, ob a- mycin-, and gen amycin-imp egna ed bone cemen s may also induce AKI (Cu is e al. 2005, Pa ick e al. 2006, Lau and Kuma 2013, Johansson e al. 2016). In ou s udy, aminogly- coside bone cemen (94% gen amycin) was used in all pa ien s and s ill we had a low a e o AKI. Fu he mo e, AKI a es we e simila be ween cemen less, hyb id, and ully cemen ed a h oplas ies. Ou s udy alida es ASA class as an independen isk ac o o AKI in a h oplas y pa ien s. This is in acco dance wi h coho s comp ising pa ien s unde going a la ge a ie y o o hopedic ope a ions, including a ious a h oplas ies, ac- u es, and os eo omies (Bell e al. 2015). P e ious s udies on a h oplas y pa ien s (Ja a i e al. 2010, Kimmel e al. 2014) again included mul iple mo bidi ies in he mul i a iable model alongside he ASA class, which migh explain why ASA class was no a signi i can p edic o o AKI in hese s udies. Ou odds a ios in di e en ASA classes we e high, and we sugges ha eade s should no in e p e hese esul s as ela i e isk. Table 3. Mul i a iable eg ession esul s Desc ip o OR (95% CI) p- alue Du a ion o ope a ion a 1.1 (1.0–1.2) 0.003 ASA classi i ca ion 1 and 2 Re e ence 3 4.4 (1.8–11) 0.002 4 13 (3.7–46) < 0.001 Bila e ali y Unila e al Re e ence Bila e al 0.31 (0.06–1.6) 0.2 Body mass index 1.1 (1.0–1.1) 0.02 Fixa ion echnique Cemen less Re e ence Hyb id 1.7 (0.56–5.1) 0.3 To al cemen 1.2 (0.39–3.7) 0.8 Join Hip Re Knee 1.0 (0.43–2.5) 1 Ope a ion ype P ima y Re e ence Re ision 0.46 (0.12–1.7) 0.2 P eope a i e eGFR b 0.98 (0.97–1.0) 0.03 a Du a ion in 10-min in e als ( om incision o end o wound closu e). b eGFR calcula ed using CKD-EPI o mula. Table 4. Fac o s possibly con ibu ing o de elopmen o AKI in 58 pa ien s Risk ac o F equency (missing) ASA class ≥ 3 47 (2) Baseline BP ≤ 80 mmHg a 14 (4) Blood ans usion 14 (5) BMI ≥ 30 26 (12) Du a ion ≥ 120 min 30 (2) eGFR ≤ 90 mL/min 49 Medica ion combina ion b 10 (9) Pe iope a i e NSAIDs 20 (9) P eope a i e anemia c 11 Use o asoac i es d 22 (4) a Pe iope a i e blood p essu e. b Concomi an use o 3 o mo e o he ollowing d ugs in pe iope a- i e pe iod: diu e ics, angio ensin-con e ing enzyme inhibi o s (ACEis) o angio ensin ecep o blocke s (ARBs), NSAIDs. c Anemia was de i ned as hemoglobin < 117 g/L in women and < 134 g/L in men. d A opine o e ile ine. Figu e 2. Su i al cu e, wi h all-cause mo ali y as endpoin . 10927 Ja msa D.indd 37410927 Ja msa D.indd 374 6/20/2017 5:25:15 PM6/20/2017 5:25:15 PM Ac a O hopaedica 2017; 88 (4): 370–376 375 We a e he i s g oup o i nd ha du a ion o ope a ion was a s ong and independen isk ac o o AKI. O he s udies ha ha e ound an insigni i can associa ion be ween du a ion o ope a ion and AKI we e case-con ol s udies wi h smalle con- ol g oups (Ja a i e al. 2010, Kimmel e al. 2014) o smalle coho s udies (Weinga en e al. 2012). In ou s udy popula- ion, he incidence o AKI almos doubled when he du a ion o su ge y exceeded 120 min. Clinical expe ience shows ha p olonged su ge y is usually caused by pe iope a i e p ob- lems, and longe du a ion has plen y o associa ions wi h o he pe iope a i e ac o s ( o example, blood ans usions, cooling, coagulopa hies, and use o anes he ics) ha could media e AKI. Howe e , he sys emic in l amma ion caused by inc eased issue damage due o he p olonged du a ion o he ope a ion and p olonged ou nique use could also be he explana ion (And es e al. 2003, Basile e al. 2012). Conce ning an ibio ic p ophylaxis, ea lie s udies ha e shown an associa ion be ween gen amycin and AKI (Ross e al. 2013, Bell e al. 2014, C ax o d e al. 2014, Johansson e al. 2016). In ou coho , gen amycin was a ely used. Ins ead o gen amycin, he seconda y op ion o ce u oxime was clindamycin. Howe e , in he AKI pa ien s, only 3 pa ien s ecei ed clindamycin, so signi i can esul s in uni a iable anal- ysis conce ning clindamycin should be iewed wi h cau ion. Whene e AKI occu ed, i s ou come was good—wi h only 2 o he 58 AKI pa ien s ecei ing dialysis, and nei he o hem o mo e han 4 weeks. This esul is in acco dance wi h he ea lie li e a u e. Ja a i e al. (2010) epo ed ha 7 ou o 98 o hei AKI pa ien s ecei ed dialysis, whe eas Kimmel e al. (2014) epo ed ha none o hei AKI pa ien s (AKI s age I o F, n = 22) ecei ed dialysis pos ope a i ely. Al hough esolu ion o kidney unc ion was common, he su i al o pa ien s wi h AKI was su p isingly poo , which wa an s u he esea ch. In hal o he cases, no speci i c cause o condi ion unde ly- ing he AKI was ound. I is possible ha a igge ing ac o o AKI in hese pa ien s migh simply be he sys emic s ess caused by he ope a ion. Pos - o p eope a i e in ec ion, pa a- ly ic ileus, and ca diac causes we e ema kable ac o s unde - lying AKI and se ed o explain almos hal o he AKI cases. By p e en ing hese ac o s, i migh be possible o educe a es o AKI. We do, howe e , acknowledge ha in some cases he ac o s leading o AKI could no be iden i i ed because o he e ospec i e na u e o he s udy. The s eng h o he p esen s udy was he ela i ely la ge pa ien coho compa ed o p e ious s udies. We used a ull, unselec ed coho as con ols in ou analysis. Mo eo e , he unde lying causes o AKI ha e no been adequa ely epo ed be o e. The s udy also had some limi a ions. Because he eco ds o some pa ien s we e no a ailable, he e was a lack o in o ma ion on he unde lying easons o hei AKI. As SC was measu ed o clinical indica ions and no ou inely in all pa ien s, i is p obable ha some AKI cases we e no iden i- i ed. Thus, i is likely ha mo e AKI cases would ha e been ound i he SC o all pa ien s had been sc eened e e y day du ing he i s pos ope a i e week. In he ea lie li e a u e, no all pos ope a i e SC le els we e eco ded ei he (Ja a i e al. 2010), so he esul s a e compa able. Kimmel e al. (2014) had pos ope a i e SC measu ed om e e y pa ien and epo ed ema kably highe a es o AKI han ou s (52 pe 1,000 s. 3.3 pe 1,000). In ou s udy popula ion, he incidence o AKI in pa ien s o whom pos ope a i e SC was measu ed was much smalle (10 pe 1,000 ope a ions), al hough his incidence numbe is p obably o e s a ed due o selec ion bias. In summa y, we ound ha high a es o AKI migh be p e- en ed i he isk p o i le o pa ien s is a o able. Con a y o ou hypo hesis, we ound e y low a es o AKI. The esul s sugges ha he isk o de eloping AKI is ele a ed in pa ien s wi h a du a ion o ope a ion exceeding 120 min, highe BMI, ASA class ≥ 3, o impai ed kidney unc ion p eope a i ely (eGFR< 90 mL/min). We he e o e sugges ca e ul enal moni o ing pos ope a i ely o pa ien s wi h hese isk ac o s, in o de o iden i y AKI. I is impo an o i nd and op imize pa ien s who a e a isk o AKI al eady p eope a i ely, o p e- en u u e kidney mani es a ions—and also o communica e he isk o he pa ien when conside ing a p ocedu e aimed a imp o ing quali y o li e in a high- isk pa ien . Supplemen a y da a A di ec ed acyclic g aph is a ailable as supple- men a y da a in he online e sion o his a icle, h p://dx.doi.o g/10.1080/17453674.2017.1301743. All he au ho s con ibu ed o he design o he s udy. PJ collec ed he da a. PJ, EJ, and LPL con ibu ed o he edi ing o he da a. PJ w o e he d a o he manusc ip and e ised he manusc ip acco ding o he co ec ions made by he o he au ho s. And es B M, Taub D D, Gu kan I, Wenz J F. Pos ope a i e e e a e o al knee a h oplas y: he ole o cy okines. Clin O hop Rela Res 2003; (415): 221-31. A eline C, Le oux A, Vau ie P, Cogne F, Le He e H, Bonne F. Risk ac o s o enal dys unc ion a e o al hip a h oplas y. Ann F Anes h Reanim 2009; 28(9): 728-34. Basile D P, Ande son M D, Su on T A. Pa hophysiology o acu e kidney inju y. Comp Physiol 2012; 2(2), 1303-53. Bell S, Da ey P, Na hwani D, Ma wick C, Vadi eloo T, Sneddon J, Pa on A, Bennie M, Fleming S, Donnan PT. Risk o AKI wi h gen amicin as su gical p ophylaxis. J Am Soc Neph ol 2014; 25(11): 2625-32. Bell S, Dekke F W, Vadi eloo T, Ma wick C, Deshmukh H, Donnan P T, Van Diepen M. Risk o pos ope a i e acu e kidney inju y in pa ien s unde going o hopaedic su ge y-de elopmen and alida ion o a isk sco e and e ec o acu e kidney inju y on su i al: obse a ional coho s udy. BMJ 2015; 11; 351: h5639. Bellomo R, Kellum J A, Ronco C. De i ning and classi ying acu e enal ail- u e: om ad ocacy o consensus and alida ion o he RIFLE c i e ia. In ensi e Ca e Med 2007; 33(3): 409-13. Coca S G, Singanamala S, Pa ikh C R. Ch onic kidney disease a e acu e kidney inju y: a sys ema ic e iew and me a-analysis. Kidney In 2012; 81(5): 442-8. Cu is J M, S e nhagen V, Ba s D. Acu e enal ailu e a e placemen o ob amycin-imp egna ed bone cemen in an in ec ed o al knee a h o- plas y. Pha maco he apy 2005; 25(6): 876-80. 10927 Ja msa D.indd 37510927 Ja msa D.indd 375 6/20/2017 5:25:15 PM6/20/2017 5:25:15 PM 376 Ac a O hopaedica 2017; 88 (4): 370–376 C ax o d S, Bayley E, Needo M. An ibio ic-associa ed complica ions ol- lowing lowe limb a h oplas y: a compa ison o wo p ophylac ic egimes. Eu J O hop Su g T auma ol 2014; 24(4): 539-43. Fou nie J P, Somme A, Du ieu G, Pou ain J C, Lapey e-Mes e M, Mon a- s uc J L. Mo e on he “T iple Whammy”: an ihype ensi e d ugs, non-s e- oidal an i-in l amma o y agen s and acu e kidney inju y - a case/non-case s udy in he F ench pha maco igilance da abase. Ren Fail 2014; 36(7): 1166-8. Ja a i S M, Huang R, Joshi A, Pa izi J, Hozack W J. Renal impai men ol- lowing o al join a h oplas y: who is a isk? J A h oplas y 2010; 25(6 Suppl): 49-53. Johansson S, Ch is ensen O M, Tho sma k A H. A e ospec i e s udy o acu e kidney inju y in hip a h oplas y pa ien s ecei ing gen amicin and dicloxacillin. Ac a O hop 2016; 87(6): 589-91. Kimmel L A, Wilson S, Jana dan J D, Liew S M, Walke R G. Incidence o acu e kidney inju y ollowing o al join a h oplas y: a e ospec i e e iew by RIFLE c i e ia. Clin Kidney J 2014; 7(6): 546-51. La ance J P, Mille D R. Acu e kidney inju y associa es wi h inc eased long- e m mo ali y. J Am Soc Neph ol 2010; 21(2): 345-52. Lau B P, Kuma V P. Acu e kidney inju y (AKI) wi h he use o an ibio ic- imp egna ed bone cemen in p ima y o al knee a h oplas y. Ann Acad Med Singapo e 2013; 42(12): 692-5. Lee A, Coope M G, C aiq J C, Knigh J F, Keneally J P. E ec s o nons e- oidal an i-in l amma o y d ugs on pos ope a i e enal unc ion in adul s wi h no mal enal unc ion. Coch ane Da abase Sys Re 2007; 18;(2): CD002765. Le ey A S, S e ens L A, Schmid C H, Zhang Y L, Cas o A F 3 d, Feldman H I, Kusek J W, Egge s P, Van Len e F, G eene T, Co esh J. A new equa ion o es ima e glome ula i l a ion a e. Ann In e n Med 2009; 150(9): 604-12. Nielson E, Henn ikus E, Lehman E, Me s B. Angio ensin axis blockade, hypo ension, and acu e kidney inju y in elec i e majo o hopedic su ge y. J Hosp Med 2014; 9(5): 283-8. Pa ick B N, Ri ey M P, Alling on D R. Acu e enal ailu e associa ed wi h ancomycin- and ob amycin-laden cemen in o al hip a h oplas y. Ann Pha maco he 2006; 40(11): 2037-42. Pe egaa d H, Damhol M B, Solgaa d S, Pe e sen M B. Renal unc ion a e elec i e o al hip eplacemen - Incidence o acu e kidney inju y and p e a- lence o ch onic kidney disease. Ac a O hop 2016; 87(3): 235-8. Ross A D, Boscainos P J, Malhas A, Wigde owi z C. Pe i-ope a i e enal mo - bidi y seconda y o gen amicin and l ucloxacillin chemop ophylaxis o hip and knee a h oplas y. Sco Med J 2013; 58(4): 209-12. Tex o J, Ha d J, Knüppel S. DAGi y: A g aphical ool o analyzing causal diag ams. Epidemiology 2011; 22(4): 745. Weinga en T N, Gu ie i C, Ja e P D, B own D R, Be n son N J, Cala o R D J , Ko D J, Be y D J, Ga o ic V D, Nicholson W T, Sch oede D R, Sp ung J. Acu e kidney inju y ollowing o al join a h oplas y: e ospec- i e analysis. Can J Anaes h 2012; 59(12): 1111-8. Ympa Y P, Sak Y, Reinha K, Vincen J L. Has mo ali y om acu e enal ailu e dec eased? A sys ema ic e iew o he li e a u e. Am J Med 2005; 118(8): 827-32 10927 Ja msa D.indd 37610927 Ja msa D.indd 376 6/20/2017 5:25:15 PM6/20/2017 5:25:15 PM