Mortality of patients with multiple sclerosis: a cohort study in UK primary care

Transcription

1 Mortlity of ptients with multiple sclerosis: cohort study in UK primry cre The Hrvrd community hs mde this rticle openly vilble. Plese shre how this ccess benefits you. Your story mtters Cittion Jick, S. S., L. Li, G. J. Flcone, Z. P. Vssilev, nd M.-A. Wllnder Mortlity of ptients with multiple sclerosis: cohort study in UK primry cre. Journl of Neurology 261 (8): doi: /s s Published Version doi: /s Citble link Terms of Use This rticle ws downloded from Hrvrd University s DASH repository, nd is mde vilble under the terms nd conditions pplicble to Other Posted Mteril, s set forth t nrs.hrvrd.edu/urn-3:hul.instrepos:dsh.current.terms-ofuse#laa

2 J Neurol (2014) 261: DOI /s ORIGINAL COMMUNICATION Mortlity of ptients with multiple sclerosis: cohort study in UK primry cre S. S. Jick L. Li G. J. Flcone Z. P. Vssilev M.-A. Wllnder Received: 23 Mrch 2014 / Accepted: 6 My 2014 / Published online: 18 My 2014 Ó The Author(s) This rticle is published with open ccess t Springerlink.com Abstrct We imed to estimte rtes, cuses nd risk fctors of ll-cuse mortlity in lrge popultion-bsed cohort of multiple sclerosis (MS) ptients compred with ptients without MS. Using dt from the UK Generl Prctice Reserch Dtbse, we identified MS cses dignosed during nd vlidted using ptients originl records where possible. We lso included MS cses during identified nd vlidted in n erlier study. Cses were mtched to up to ten referents without MS by ge, sex, index dte (dte of first MS dignosis for cses nd equivlent reference dte for controls), generl prctice nd length of medicl history before first MS dignosis. Ptients were followed up to identify deths; hzrd rtios (HRs) nd 95 % confidence intervls Electronic supplementry mteril The online version of this rticle (doi: /s ) contins supplementry mteril, which is vilble to uthorized users. S. S. Jick (&) L. Li Boston Collbortive Drug Surveillnce Progrm, Boston University School of Public Helth, 11 Muzzey Street, Lexington, MA 02421, USA e-mil: sjick@bu.edu G. J. Flcone Deprtment of Neurology, Msschusetts Generl Hospitl, Boston, MA, USA G. J. Flcone Deprtment of Epidemiology, Hrvrd School of Public Helth, Boston, MA, USA Z. P. Vssilev Byer HelthCre Phrmceuticls, Whippny, NJ, USA M.-A. Wllnder Deprtment of Public Helth nd Cring Science, Uppsl University, Uppsl, Sweden (CIs) were estimted using Cox-proportionl regression. MS ptients (N = 1,822) hd significntly incresed risk of ll-cuse mortlity compred with referents (N = 18,211); djusted HR 1.7 (95 % CI ). Compred with referents, femle MS ptients hd higher but not significntly different HR for deth thn mles; djusted HR 1.86 (95 % CI ) vs. HR 1.31 (95 % CI ), respectively. The most commonly recorded cuse of deth in MS ptients ws MS (41 %), with higher proportion recorded mong younger ptients. A significntly higher proportion of referents thn MS ptients hd cncer recorded s cuse of deth (40 vs. 19 %). Ptients with MS hve significnt 1.7-fold incresed risk of ll-cuse mortlity compred with the generl popultion. MS is the most commonly recorded cuse of deth mong MS ptients. Keywords Multiple sclerosis Epidemiology Cuse of deth Cohort nlysis Mortlity Introduction Multiple sclerosis (MS) is chronic, progressive neurologicl disorder, nd is the mjor cuse of non-trumtic disbility in young dults [1]. Mortlity rtes re significntly higher in people with MS compred with the generl popultion [2 4], yet cuses of deth nd fctors influencing survivl in MS ptients re not well understood, nd further dt ddressing these questions re needed. One importnt question is to wht extent cn the excess in mortlity rtes observed in MS ptients be ttributed to the disese. Another importnt question is wht re the risk fctors for the disese nd wht influence do they hve on mortlity. Identifying risk fctors, nd understnding their

3 J Neurol (2014) 261: reltive weight on the risk of mortlity, is vitl for improving our bilities to predict survivl nd for implementing blnced therpeutic strtegy. We therefore imed to estimte mortlity rtes, describe cuse-specific mortlity nd identify risk fctors for mortlity in lrge popultion-bsed cohort of MS ptients compred with ptients without MS in the United Kingdom (UK). Methods Dt source We conducted cohort study using prospectively collected dt from the UK Generl Prctice Reserch Dtbse (GPRD). The GPRD comprises the nonymized computerized medicl records of round 6 % of the UK popultion t ny one time [5, 6]. The dtbse is the core of the primry cre dt mnged by the Clinicl Prctice Reserch Dtlink (CPRD), nd contins informtion on ptients demogrphics, clinicl events, nd detils of specilist referrls, hospitl dmissions nd prescriptions. Additionl informtion cn be dded s free text. Prescriptions re issued directly from the GPs computer, hence ll re recorded. Ptients in the GPRD re representtive of the UK popultion with regrd to ge, sex nd geogrphicl distribution. Vlidtion studies hve demonstrted the qulity of the dt in the GPRD to be generlly high [5], nd the dtbse hs been used for severl studies on MS [7 15]. The study protocol ws pproved by the CPRD Independent Scientific Advisory Committee for UK Medicines nd Helthcre products Regultory Agency dtbse reserch. Study design nd MS cse scertinment We identified cohort of incident nd prevlent MS ptients in the GPRD between Jnury 1993 nd December This ws two-step process involving two study periods. Firstly, we identified ll possible nd probble incident MS cses, nd ll prevlent MS cses occurring between Jnury 2001 nd December Secondly, we dded to this cse set, ll incident nd prevlent MS cses identified nd vlidted in our erlier studies conducted on the GPRD (between 1993 nd 2000) [7, 8] from prctices tht were still contributing to the dtbse in To identify incident MS cses between 2001 nd 2006, we identified ptients registered in the GPRD in this study period who hd t lest two continuous yers of registrtion. Ptients with first recorded dignosis for MS were identified s potentil incident cses, nd their computerized medicl records were reviewed. Potentil incident cses were clssified s either probble, possible, or unlikely MS using n overll clinicl impression obtined fter review by neurologist (GJF) of ll vilble dt on dignoses, tretments nd referrls recorded in the ptient record: probble cse, multiple entries of MS dignosis plus tretment nd/or symptom codes; possible cse, t lest one entry of MS dignosis plus codes for tretments or symptoms tht my hve been relted to MS or multiple MS dignoses with no supporting tretment or symptoms codes; unlikely cse, one entry of MS dignosis only nd no supporting tretment or symptoms codes, or the record contined n lternte dignosis such s myotrophic lterl sclerosis. Where vilble, ptients originl clinicl pper records contining clinicl notes, consulttion letters, specilist referrls, test results nd hospitl dischrge letters were retrieved, nd were reviewed by neurologist (GJF) to vlidte the MS dignosis nd clssify the cse. Ech cse ws clssified ccording to the McDonld criteri for MS [16] nd ctegorized s definite, possible or unlikely. For cses where GP comments strongly indicted tht the ptient hd MS but objective dt were unvilble to fulfill the McDonld criteri, fourth ctegory clled probble ws used. Where there ws uncertinty, second neurologist reviewed the ptient records nd finl decision ws reched by consensus. The index dte ws the dte of MS dignosis. Comprison group selection Ech cse ws rndomly mtched to up to 10 referent subjects without recorded dignosis of MS t ny time in the GPRD by yer of birth, sex, index dte, generl prctice nd length of medicl history recorded before the index dte. The index dte for referents ws the dte of dignosis of their mtched MS cse. Follow-up Ptients were followed up from the index dte to identify deths. The end of follow-up ws the dte of ptient s deth, the dte of trnsfer out of the prctice, or the dte of lst dt collection in the dtbse (31 July 2012), whichever cme first. The cuse of deth ws scertined from electronic nd/or originl clinicl pper records. To supplement informtion on cuse of deth recorded in the GP records, we were ble to link to deth registry dt for ptients in prctices in Englnd, nd we requested dt from dditionl electronic GP notes where the cuse of deth ws unknown, but there ws n indiction tht dditionl electronic GP notes were vilble.

4 1510 J Neurol (2014) 261: Covrite informtion Informtion on the following vribles ws extrcted from the dtbse: ge t first MS dignosis, sex, nd lifestyle fctors closest to nd before the first MS dignosis including smoking, body mss index (BMI) nd lcohol buse. Informtion on chronic comorbidities (recorded t ny time in the dtbse), cute illnesses (recorded t or within 1 yer before dignosis or ny time fter), nd comedictions (t or within 6 months before dignosis or ny time fter) were lso extrcted. Comorbidities evluted included both chronic diseses (chronic obstructive pulmonry disese or sthm, depression, dibetes, hypertension, hert disese nd cncer) nd cute illnesses [infections (respirtory, pneumoni/influenz, urinry trct, skin, eye or er, or other), pin (joint, spine, muscle, migrine, eye or er) nd dyspepsi]. Comedictions evluted included ntibiotics, ntidepressnts, skeletl muscle relxnts, ntipsychotics, nti-prkinson s drugs, nticonvulsnts, opioids, non-steroidl nti-inflmmtory drugs (NSAIDs), topicl NSAIDs, other nlgesics, proton pump inhibitors, steroids, topicl steroids, sttins, nd orl contrceptives. We lso clculted the Chrlson Comorbidity Index [17] before the first recorded MS dignosis. We were unble to obtin informtion on use of interferon bet becuse, in the UK, it is mostly prescribed in secondry cre nd is not lwys cptured in GP records. For the 902 MS ptients with originl clinicl records, we retrieved dditionl informtion on MS subtype [relpsing-remitting MS (RRMS), primry progressive MS (PPMS), secondry progressive MS (SPMS) or unknown], nd MS symptoms t onset. These were clssified into four min groups: sensory, motor, optic neuritis nd other. The ltter group included other optic nd visul nomlies, other crnil nerve nomlies, dysrthri nd stheni. Ptients without symptoms recorded were treted s seprte group with missing symptoms. Sttisticl nlyses All definite, probble nd possible MS cses were included in the nlyses. Ptient chrcteristics were described t or before the index dte nd seprtely t the end of follow-up using counts [percentges (%)] for discrete nd ctegoricl vribles, nd mens [stndrd devition (SD)] for continuous vribles. We compred ptients chrcteristics including lifestyle risk fctors, comorbidities nd comedictions between the MS ptients nd referents. For continuous vribles, we used t tests; for ctegoricl vribles, we used Chi squre tests or Fisher s exct tests where necessry. Crude deth rtes with 95 % confidence intervls (CIs) were clculted overll nd strtified by ge t first dignosis, sex nd type of MS. Cuse of deth ws described where possible. Survivl probbilities for fixed ctegoricl vribles relted to ll-cuse mortlity were estimted using Kpln Meier survivl nlyses, both overll nd strtified by ge t first dignosis nd sex. Hzrd rtios (HRs) nd 95 % CIs for ll-cuse mortlity were estimted using Coxproportionl regression models djusted for potentil confounding vribles. Sttisticl nlyses were performed using SAS version 9.2 (SAS Institute, Cry, NC). Results Between Jnury 2001 nd December 2006, 1278 incident nd 63 prevlent MS cses were identified. In ddition, 435 incident nd 46 prevlent MS cses identified between 1993 nd 2000 were retrieved from previous studies [7 10, 12 14], giving totl of 1,822 MS cses (1,507 definite or probble nd 315 possible), mtched to 18,211 referents. Nerly, three qurters of MS cses were femle nd the men ge t dignosis ws 42.1 yers. PPMS ptients were generlly older t dignosis thn RRMS ptients (men ge 50 vs. 40 yers; p \ 0.001), nd more likely to be mle (42.4 vs %; p \ 0.001) nd current smokers (32 vs %, p = 0.30) (Tble 1). The length of follow-up postcohort entry dte ws similr for MS ptients nd their non- MS comprtors: 7.9 yers (rnge 1 dy 19 yers). Chrcteristics of MS vs. referent subjects There were number of significnt differences between the chrcteristics, comorbidities nd comedictions of MS ptients nd their mtched referents (Tbles 1, 2, 3). At the index dte, MS ptients were more likely to be current smokers (p \ 0.05). Within the yer before the index dte, MS ptients overll were more likely to hve hd urinry trct infection (p \ 0.05), PPMS cses were more likely to hve hd n cute respirtory infection (p \ 0.05) nd RRMS cses were more likely to hve hd other infections (p \ 0.05). Both RRMS nd PPMS ptients were more likely to hve history of depression (p \ 0.05), nd to hve received, ntidepressnts, nticonvulsnts, opioids, muscle relxnts nd nti-prkinson s drugs t or within the 6 months before the index dte (p \ 0.05 for ll). In ddition, during this time period, MS ptients overll were more likely to hve received sttins, NSAIDs, systemic glucocorticoids, ntibiotics nd ntipsychotics (p \ 0.05 for ll). Compred with RRMS ptients, PPMS ptients were more likely to receive symptomtic MS tretment (e.g., muscle relxnts; 16.0 vs. 6.5 %; p = ), but were less likely to receive ntibiotics (12.8 vs %; p = ). Compred with referents, t the end of follow-up MS ptients were more likely to be smokers (26.3 vs %)

5 J Neurol (2014) 261: Tble 1 Bsic chrcteristics of MS cses nd mtched referent subjects t cohort entry, overll nd strtified by type of MS Chrcteristic MS cses N = 1,822 N = 18,211 RRMS N = 769 N = 7,690 PPMS N = 125 N = 1,250 Men ge (yers) t cohort entry (index (11.79) 42.0 (11.72) (11.14) (11.15) (10.43) (10.39) dte; SD) Sex Mles 481 (26.40) 4,801 (26.36) 193 (25.10) 1,930 (25.10) 53 (42.40) 530 (42.40) Femles 1,341 (73.60) 13,410 (73.64) 576 (74.90) 5,760 (74.90) 72 (57.60) 720 (57.60) Smoking sttus Current*, 569 (31.23) 4,363 (23.96) 226 (29.39) 1,902 (24.73) 40 (32.00) 289 (23.12) Former 239 (13.12) 2,390 (13.12) 101 (13.13) 963 (12.52) 23 (18.40) 199 (15.92) Never 777 (42.65) 8,921 (48.99) 349 (45.38) 3,807 (49.51) 50 (40.00) 595 (47.60) Unknown 237 (13.01) 2,537 (13.93) 93 (12.09) 1,018 (13.24) 12 (9.60) 167 (13.36) BMI (kg/m 2 ) \ (2.52) 379 (2.08) 17 (2.21) 179 (2.33) 2 (1.60) 18 (1.44) (38.75) 6,948 (38.15) 301 (39.14) 2,980 (38.75) 46 (36.80) 408 (32.64) (23.38) 4,100 (22.51) 187 (24.32) 1,676 (21.79) 39 (31.20) 339 (27.12) C (13.61) 2,406 (13.21) 112 (14.56) 993 (12.91) 12 (9.60) 198 (15.84) Unknown 396 (21.73) 4,378 (24.04) 152 (19.77) 1,862 (24.21) 26 (20.80) 287 (22.96) Alcohol buse 25 (1.37) 370 (2.03) 3 (0.39) 133 (1.73) 5 (4.0) 50 (4.0) Men length of recorded medicl history (yers; SD) Before index dte 7.86 (4.44) 7.97 (4.43) 7.93 (4.50) 8.03 (4.49) 8.83 (4.51) 9.02 (4.52) Follow-up fter index dte 7.85 (4.50) 7.95 (4.49) 8.95 (4.33) 8.18 (4.40) 8.15 (4.32) 8.15 (4.08) For the 894 MS ptients whose type of MS ws determined through originl clinicl records BMI body mss index, MS multiple sclerosis, PPMS primry progressive MS, RRMS relpsing-remitting MS, SD stndrd devition * p \ 0.05 for comprison between ptients with MS nd mtched referent subjects (for ll MS ptients) p \ 0.05 for comprison between ptients with RRMS nd mtched referent subjects Includes 64 ptients with secondry progressive MS to hve BMI \18.5 kg/m 2 (3.6 vs. 2.0 %), history of depression (45.7 vs %), more infections, including pneumoni nd influenz (1.8 vs. 0.9 %), urinry trct infections (10.5 vs. 3.9 %), nd skin infections (11.9 vs. 9.8 %), nd to hve received more comedictions in the prior 6 months (12.3 vs. 8.5 %) (p \ 0.05 for ll) (Online Resource 1). Mortlity rtes Of the 1,822 MS cses, 130 (7.1 %) died during 14,295 person-yers of follow-up, while 573 (3.1 %) referents died during 144,760 person-yers of follow-up. The crude deth rte for MS ptients ws 9.1 (95 % CI ) per 1,000 person-yers compred with 4.0 (95 % CI ) per 1,000 person-yers for the non-ms referents. Mortlity rtes were higher in MS ptients compred with their mtched referents in ech ge group nd for both mles nd femles. Ptients with PPMS hd n lmost twofold greter mortlity rte compred with RRMS ptients (11.8 vs. 5.8 per 1,000 person-yers). However, these deth rtes were similrly higher compred with their corresponding referents; 6.2 vs. 2.9 per 1,000 person-yers for PPMS nd RRMS referents, respectively) (Tble 4). Ptients ged C50 yers t dignosis hd shorter survivl thn those ged \50 yers t dignosis (Fig. 1). For the ltter group, n incresing difference in survivl between MS cses nd their mtched non-ms referents ws observed with incresing length of follow-up. Survivl probbilities were similr for both mle nd femle MS ptients, with 10-yer survivl t 90 nd 93 %, nd 15-yer survivl t 86 nd 87 % for mles nd femles, respectively. Cuse of deth Among MS ptients who died during follow-up, the most commonly recorded cuse of deth ws MS (40.8 %), followed by pneumoni (25.4 %), cncer (18.5 %), crdiovsculr disese (13.9 %) nd non-infectious respirtory diseses (10.0 %). Among decesed referents, cuse of deth ws recorded s pneumoni in 6.8 %, cncer in

6 1512 J Neurol (2014) 261: Tble 2 Comorbidities of MS cses nd mtched referent subjects t cohort entry, overll nd strtified by type of MS Chrcteristic MS cses N = 1,822 N = 18,211 RRMS cses N = 769 N = 7,690 PPMS cses N = 125 N = 1,250 Chronic comorbidities b COPD nd sthm 302 (16.58) 2,890 (15.87) 117 (15.21) 1,245 (16.19) 18 (14.40) 197 (15.76) Depression*,, 508 (27.88) 3,677 (20.19) 188 (24.45) 1,529 (19.88) 45 (36.00) 238 (19.04) Dibetes 35 (1.92) 390 (2.14) 13 (1.69) 141 (1.83) 3 (2.40) 44 (3.52) Hypertension 137 (7.52) 1,544 (8.48) 47 (6.11) 538 (7.00) 13 (10.40) 179 (14.32) Hert disese 34 (1.87) 439 (2.41) 8 (1.04) 119 (1.55) 4 (3.20) 64 (5.12) Cncer 50 (2.74) 558 (3.06) 15 (1.95) 223 (2.90) 5 (4.00) 50 (4.00) Acute comorbidities c Acute respirtory infection 252 (13.83) 2,673 (14.68) 113 (14.69) 1,143 (14.86) 9 (7.20) 171 (13.68) Pneumoni nd influenz 18 (0.99) 169 (0.93) 11 (1.43) 70 (0.91) 1 (0.80) 14 (1.12) Urinry trct infection* 99 (5.43) 683 (3.75) 36 (4.68) 284 (3.69) 8 (6.40) 41 (3.28) Skin infection 164 (9.00) 1,566 (8.60) 68 (8.84) 652 (8.48) 5 (4.00) 96 (7.68) Eye or Er infection 4 (0.22) 33 (0.18) 3 (0.39) 15 (0.20) 0 ( ) 0 ( ) Other infection*, 149 (8.18) 1,256 (6.90) 69 (8.97) 533 (6.93) 5 (4.00) 73 (5.84) Dyspepsi 35 (1.92) 282 (1.55) 11 (1.43) 120 (1.56) 2 (1.60) 23 (1.84) Chrlson Comorbidity Index t cohort entry Low (0) 1,393 (76.45) 14,192 (77.93) 615 (79.97) 6,048 (78.65) 96 (76.80) 931 (74.48) Medium (1 2) 405 (22.23) 3,822 (20.99) 147 (19.12) 1,574 (20.47) 26 (20.80) 296 (23.68) High ([2) 24 (1.32) 197 (1.08) 7 (0.91) 68 (0.88) 3 (2.40) 23 (1.84) For the 894 MS ptients whose type of MS ws determined through originl clinicl records COPD chronic obstructive pulmonry disorder, MS multiple sclerosis, PPMS primry progressive MS, RRMS relpsing-remitting MS * p \ 0.05 for comprison between ptients with MS nd mtched referent subjects (for ll MS ptients) p \ 0.05 for comprison between ptients with RRMS nd mtched referent subjects b c p \ 0.05 for comprison between ptients with PPMS nd mtched referent subjects Includes 64 ptients with secondry progressive MS Ever before, or t cohort entry During the yer before, or t index dte 39.8 % nd crdiovsculr disese in 19.4 %. The higher proportion of cncer deths mong referents compred with the MS ptients (p \ ) is noteworthy. No mteril difference ws found in the proportion of deths recorded s suicide or ccident between MS ptients nd referents; 1.5 nd 1.9 %, respectively. However, these proportions re bsed on only two MS ptients (one suicide) nd 11 referents (two suicides). Additionl dt on cuse of deth re provided in Online Resources 2 nd 3. Among decesed MS ptients, more mles thn femles hd their cuse of deth recorded s crdiovsculr disese (24.4 vs. 9.0 %, p = 0.02) or MS (43.9 vs %, p = 0.62), while more femles thn mles hd cncer recorded s cuse of deth (20.2 vs %, p = 0.45). Ptients older t dignosis (C50 yers) hd higher proportion of deths recorded s due to cncer compred with ptients dignosed t younger ge (\50 yers) (22.4 vs %, p = 0.23), while ptients younger t dignosis hd higher proportion of deths recorded s due to MS (47.6 vs %, p = 0.12) or pneumoni (31.8 vs %, p = 0.11). Risk fctors for mortlity Compred with referents, MS ptients hd significntly incresed risk of ll-cuse mortlity; djusted HR 1.68 (95 % CI ) (Tble 5). Age ws strong effect modifier, with the youngest MS ptients yielding the highest djusted HR 13.2 (95 % CI ) for ptients ged \30 yers t dignosis. Compred with referents, femle MS ptients hd higher overll HR for deth thn mle MS ptients, lthough the HRs were not significntly different; djusted HR 1.86 (95 % CI ) vs. HR 1.31 (95 % CI ). While we observed significntly higher HR for RRMS ptients compred to referents [djusted HR 1.50 (95 % CI )], significntly

7 J Neurol (2014) 261: Tble 3 Comedictions of MS cses nd mtched referent subjects t cohort entry, overll nd strtified by type of MS Chrcteristic All MS cses N = 1,822 N = 18,211 RRMS cses N = 769 N = 7,690 PPMS cses N = 125 N = 1, (3.75) 8 (6.40) 52 (4.16) Comedictions (t cohort entry [index dte] or within the 6 months before) Systemic 180 (9.88) 713 (3.92) 69 (8.97) glucocorticoids*, Antidepressnts*,, 380 (20.86) 1,582 (8.69) 132 (17.17) 662 (8.61) 28 (22.40) 102 (8.16) Anticonvulsnts*,, 122 (6.70) 225 (1.24) 40 (5.20) 88 (1.14) 11 (8.80) 25 (2.00) Antidibetics 24 (1.32) 283 (1.55) 12 (1.56) 112 (1.46) 0 ( ) 30 (2.40) Opioids*,, 345 (18.94) 1,585 (8.70) 122 (15.86) 621 (8.08) 23 (18.40) 133 (10.64) NSAIDs*, 316 (17.34) 1,806 (9.92) 122 (15.86) 710 (9.23) 21 (16.80) 152 (12.16) Sttins* 61 (3.35) 447 (2.45) 20 (2.60) 137 (1.78) 4 (3.20) 75 (6.00) Antibiotics*, 403 (22.12) 3,598 (19.76) 179 (23.28) 1,546 (20.10) 16 (12.80) 227 (18.16) Muscle relxnts*,, 173 (9.50) 399 (2.19) 50 (6.50) 153 (1.99) 20 (16.00) 37 (2.96) Antipsychotics*, 122 (6.70) 387 (2.13) 54 (7.02) 160 (2.08) 5 (4.00) 20 (1.60) Anti-Prkinson drugs*,, 23 (1.26) 54 (0.30) 8 (1.04) 17 (0.22) 3 (2.40) 4 (0.32) PPIs* 83 (4.56) 657 (3.61) 24 (3.12) 243 (3.16) 6 (4.80) 68 (5.44) For the 894 MS ptients whose type of MS ws determined through originl clinicl records MS multiple sclerosis, NSAIDs non-steroidl nti-inflmmtory drugs, PPMS primry progressive MS, PPIs proton pump inhibitors, RRMS relpsing-remitting MS * p \ 0.05 for comprison between ptients with MS nd mtched referent subjects (for ll MS ptients) p \ 0.05 for comprison between ptients with RRMS nd mtched referent subjects p \ 0.05 for comprison between ptients with PPMS nd mtched referent subjects Includes 64 ptients with secondry progressive MS higher HR ws not found for PPMS ptients [djusted HR 1.32 (95 % CI )]. It should be noted, however, tht the number of PPMS ptients ws smll. Discussion This lrge study of 1,822 people with MS in UK primry cre provides ntionl estimtes of ll-cuse mortlity mong this ptient group. We observed 1.7-fold incresed risk of ll-cuse mortlity for MS ptients compred with the generl popultion. This increse in risk is in line with reports from other lrge popultion-bsed cohorts [2 4, 18 21], lthough direct comprisons re difficult owing to differences in geogrphicl regions nd/or study periods. Consistent with other reports, MS ws the min recorded cuse of deth (40.8 %) in MS ptients in our study. This proportion is lower thn tht in other studies, lbeit from wide rnge of geogrphicl popultions, where MS ws the reported cuse of deth in more thn 50 % of ffected ptients [4]. Respirtory diseses, infections, crdiovsculr diseses nd cncers re other commonly reported cuses of deth in ptients with MS [2, 22], nd our results re consistent with these findings. It should be noted tht the rel cuses of deth in ptients with MS re not lwys recorded in ptient records or registries. While it is ccepted tht MS is not per se ftl disese, neither our study nor prior studies hve been ble to identify ll direct cuses of deth in these ptients. Thus, complete informtion on cuse of deth remins unttinble. The lower proportion of MS ptients with cncer s the cuse of deth compred to non-ms referents could be due to incresed susceptibility to cute cuses of deth (such s infections or cute crdiovsculr events). These cute nd sometimes ftl comorbidities my ct s competing risks, truncting the long ltency periods needed for neoplstic disese to express cliniclly. Current evidence suggests tht frequency of suicides my be higher in MS ptients [23]. In only three ptients ws suicide the recorded cuse of deth in our study (one MS ptient nd two referents), which ws insufficient to detect ny mteril difference between the ptient groups. The progressive nture of MS is reflected in our study in the incresing differences in ll-cuse mortlity rtes seen between MS ptients nd their mtched non-ms referents with incresing length of follow-up. It is interesting tht ll HRs were ttenuted when risk fctors for comedictions ssocited with MS were included in the model. This suggests tht some of the incresed risk for deth is ssocited with the comorbidities nd resulting comedictions

8 1514 J Neurol (2014) 261: Tble 4 Deth rte per 1,000 person-yers in ptients with MS nd mtched referent subjects overll, strtified by ge t first dignosis of MS, sex, nd type of MS CI confidence intervl, MS multiple sclerosis, RRMS relpsing-remitting MS, PPMS primry progressive MS For the 894 MS cses confirmed by originl clinicl records b Including 64 ptients with SPMS Deths (N) Popultion size Person-yers Deth rtes (95 % CI) Overll Ptients with MS 130 1,822 14, ( ) Referent subjects , , ( ) Age group t dignosis (yers) \30 Ptients with MS , ( ) Referent subjects 5 2,666 18, ( ) Ptients with MS , ( ) Referent subjects 48 5,217 40, ( ) Ptients with MS , ( ) Referent subjects 116 5,634 47, ( ) Ptients with MS , ( ) Referent subjects 180 3,380 27, ( ) C60 Ptients with MS ( ) Referent subjects 224 1,314 10, ( ) Sex Mle Ptients with MS , ( ) Referent subjects 228 4,801 37, ( ) Femle Ptients with MS 89 1,341 10, ( ) Referent subjects , , ( ) Type of MS RRMS Ptients with MS ,b 6, ( ) Referent subjects 183 7,690 62, ( ) PPMS Ptients with MS , ( ) Referent subjects 63 1,250 10, ( ) tht occur more frequently in ptients with MS. It is notble tht the HRs for deth in the MS cohort compred to the non-ms cohort were highest in the younger ptients nd decresed with incresing ge. It is possible tht the decresed bseline risk of deth in younger people contributed to the higher HR, while bseline risks incresed for ll ptients with incresing ge leding to lower HRs. We did not, however, find ny sttisticlly significnt difference in survivl between mles nd femles, lthough the djusted HR for ll-cuse mortlity ws higher for femles. Other reports concerning gender difference in survivl hve been mixed [2, 4, 18, 19]. MS ptients in this study were more likely to hve been dignosed with depression in the period between first MS symptoms nd first MS dignosis. Dignoses of depression in the yers before first symptoms were similr in the MS ptients nd their mtched non-ms referents. Given their timing, these findings cn be interpreted either s rective in the setting of neurologicl disese of uncertin etiology, or s direct result of neurl dmge cused by the ongoing inflmmtory process ffecting the CNS. In ny cse, these results highlight tht depression, together with the hevy burden it imposes on functionl sttus, becomes relevnt problem in MS subjects long before finl dignoses of MS cn be reched. The higher incidence of urinry trct infection in the months prior to MS dignosis is consistent with decresed mobility found in MS ptients, prticulrly those with PPMS. While the incidence of Prkinson s disese ws higher in the MS group compred to the non-ms referents, the bsolute number of

9 J Neurol (2014) 261: Fig. 1 Kpln Meier plot for the survivl probbilities (llcuse mortlity) of ptients \50 yers t index dte. MS multiple sclerosis. b Kpln Meier plot for the survivl probbilities (ll-cuse mortlity) of ptients C50 yers t index dte. MS multiple sclerosis Survivl Probbility 100.0% 95.0% 90.0% 85.0% 80.0% 75.0% 70.0% Log-rnk P-vlue : <.0001 Referent subjects Ptients with MS Time since index dte (yers) b 100.0% 95.0% Survivl Probbility 90.0% 85.0% 80.0% 75.0% 70.0% Log-rnk P-vlue : <.0001 Referent subjects Ptients with MS Time since index dte (yers) cses ws smll. Previous reports hve not found similr differences. It is possible tht t lest some of the 23 cses hd Prkinsonin symptoms due to compromise of the bsl gngli by the underlying neuro-inflmmtory process, but were misdignosed nd treted s idiopthic Prkinson s disese. Alterntively, the ssocition could be explined by unmesured confounders. Strengths of our study include the lrge smple size, the high qulity of the dtbse nd the controlled study design. The representtiveness of the study popultion mens tht the results cn be generlized to the UK s whole, thus the study hs high externl vlidity. In ddition, for some MS cses, we were ble to vlidte the dignosis by ccessing ptients originl clinicl records, which long with linkge to deth registry dt, enbled dditionl clinicl informtion to be obtined. We were lso ble to obtin nd nlyze dt on lrge number of potentil risk fctors for mortlity. Our study lso hs limittions. We were unble to vlidte the MS dignosis vi originl clinicl records for ll ptients, thus it is likely tht some misclssifiction of MS cses occurred. Any misclssifiction would likely hve been rndom, non-differentil, nd would hve hd little effect on the mortlity rtes in the study. Any effects on the HRs would likely hve been smll nd bised towrd the null. However, when we repeted the min nlyses for cses whose dignosis ws mde on originl clinicl records nd for cses whose dignosis ws mde from electronic records only, we found the results from the two groups to be consistent. When we repeted the nlyses gin, restricted to definite nd probble MS ptients, the results were similr to the full group nlysis (Online

10 1516 J Neurol (2014) 261: Tble 5 Hzrd rtios nd 95 % confidence intervls for ll-cuse mortlity mong MS cses versus mtched referent subjects Chrcteristic Deth in MS ptients N = 130 Deths in non-ms referents N = 573 HR (95 % CI) Model 1 Model 2 b Overll 130 (100) 573 (100) 2.30 ( ) 1.68 ( ) Age t first MS dignosis (yers) \30 8 (6.2) 5 (0.87) ( ) ( ) (15.4) 48 (8.4) 4.27 ( ) 3.29 ( ) (26.9) 116 (20.2) 3.13 ( ) 2.16 ( ) (23.1) 180 (31.4) 1.66 ( ) 1.33 ( ) C60 37 (28.5) 224 (39.1) 1.82 ( ) 1.47 ( ) Sex Mle 41 (31.5) 228 (39.8) 1.79 ( ) 1.31 ( ) Femle 89 (68.5) 345 (60.2) 2.63 ( ) 1.86 ( ) Type of MS c RRMS d 40 (30.8) 183 e (31.9) 1.94 ( ) 1.50 ( ) PPMS 12 (9.2) 63 f (11.0) 1.89 ( ) 1.32 ( ) CI confidence intervl, HR hzrd rtio, MS multiple sclerosis, PPMS primry progressive MS, RRMS relpsing-remitting MS, SPMS secondry progressive MS Adjusted for mtching vribles (ge, sex, index dte, GP, nd length of medicl history recorded before the index dte) b Adjusted for smoking nd comedictions (including ntidepressnts, nticonvulsnts, opioids, muscle relxnts) within 6 months before or t the index dte, in ddition to the mtching vribles c MS cses confirmed by originl clinicl pper records d Including ptients with SPMS e Mtched to ptients with RRMS f Mtched to ptients with PPMS Resources 4 7). We were lso unble to identify cuse of deth for round 15 % of decesed MS ptients nd 17 % of decesed referents, which my hve influenced comprisons of cuse of deth. However, we found the most common cuses of deths were consistent with those reported elsewhere [22]. Our study ws underpowered to explore risk fctors for cuse-specific mortlity, nd it is lso possible tht there ws some underestimtion of our reported ll-cuse mortlity rtes. This is becuse followup time my not hve been long enough to observe mortlity, especilly for MS cses identified between 2001 nd It is lso possible tht smll number of deths were not recorded by the GP though recording of deth is required of ll GPs nd this number should be miniml. In ddition, we were unble to evlute the influence of interferon bet, which hs been shown to slow disese progression [24 26] nd reduce ll-cuse mortlity [26 28] in ptients with RRMS. Although this my hve ffected our estimtes of ll-cuse mortlity mong RRMS ptients, it is unlikely to hve ffected the HRs ssocited with fctors found to be predictive of ll-cuse mortlity, s such confounding by indiction is unlikely. In summry, our findings show tht ptients with MS hve incresed ll-cuse mortlity rtes compred with the generl popultion, nd tht some of this increse is likely relted to the comorbidities tht occur more frequently in ptients with MS. Most deths in MS ptients re recorded s due to the disese itself. Acknowledgments This study ws funded by Byer HelthCre AG. We thnk Susn Bromley, Independent Medicl Writer (Oxford, UK) for ssistnce in the drfting nd editing the mnuscript, funded by Byer HelthCre AG. This study is bsed on dt from the Full Feture GPRD obtined under licence from the UK Medicines nd Helthcre Products Regultory Agency. Conflicts of interest Susn Jick nd Lin Li work for the BCDSP, which hs received reserch funding from Byer HelthCre AG for this study. Guido Flcone hs no conflicts of interest. Zdrvko Vssilev is slried employee of Byer HelthCre Phrmceuticls. Mri-Ann Wllnder ws slried employee of Byer HelthCre AG t the time of the study. Ethicl stndrd The conduct of this study ws pproved by the CPRD Independent Scientific Advisory Committee for UK Medicines nd Helthcre products Regultory Agency dtbse reserch. Open Access This rticle is distributed under the terms of the Cretive Commons Attribution License which permits ny use, distribution, nd reproduction in ny medium, provided the originl uthor(s) nd the source re credited.

11 J Neurol (2014) 261: References 1. Puglitti M, Rosti G, Crton H, Riise T, Drulovic J, Vecsei L, Milnov I (2006) The epidemiology of multiple sclerosis in Europe. Eur J Neurol 13: Sclfri A, Knppertz V, Cutter G, Goodin DS, Ashton R, Ebers GC (2013) Mortlity in ptients with multiple sclerosis. Neurology 81: Herrett E, Thoms SL, Schoonen WM, Smeeth L, Hll AJ (2010) Vlidtion nd vlidity of dignoses in the Generl Prctice Reserch Dtbse: systemtic review. Br J Clin Phrmcol 69: Willims T, vn St TP, Puri S, Eton S (2012) Recent dvnces in the utility nd use of the Generl Prctice Reserch Dtbse s n exmple of UK Primry Cre Dt resource. Ther Adv Drug Sf 3: Alonso A, Jick SS, Hernn MA (2006) Allergy, histmine 1 receptor blockers, nd the risk of multiple sclerosis. Neurology 66: Alonso A, Jick SS, Jick H, Hernn MA (2006) Antibiotic use nd risk of multiple sclerosis. Am J Epidemiol 163: Alonso A, Jick SS, Olek MJ, Ascherio A, Jick H, Hernn MA (2005) Recent use of orl contrceptives nd the risk of multiple sclerosis. Arch Neurol 62: Alonso A, Jick SS, Olek MJ, Hernn MA (2007) Incidence of multiple sclerosis in the United Kingdom: findings from popultion-bsed cohort. J Neurol 254: Bzelier MT, vn St T, Uitdehg BM, Cooper C, Leufkens HG, Vestergrd P, Bentzen J, de Vries F (2011) The risk of frcture in ptients with multiple sclerosis: the UK generl prctice reserch dtbse. J Bone Miner Res 26: Hernn MA, Jick SS (2006) Heptitis B vccintion nd multiple sclerosis: the jury is still out. Phrmcoepidemiol Drug Sf 15: Hernn MA, Jick SS, Logroscino G, Olek MJ, Ascherio A, Jick H (2005) Cigrette smoking nd the progression of multiple sclerosis. Brin 128: Hernn MA, Jick SS, Olek MJ, Jick H (2004) Recombinnt heptitis B vccine nd the risk of multiple sclerosis: prospective study. Neurology 63: Mckenzie IS, Mornt SV, Bloomfield GA, McDonld TM, O Riordn J (2014) Incidence nd prevlence of multiple sclerosis in the UK : descriptive study in the Generl Prctice Reserch Dtbse. J Neurol Neurosurg Psychitry 85: Polmn CH, Reingold SC, Edn G, Filippi M, Hrtung HP, Kppos L, Lublin FD, Metz LM, McFrlnd HF, O Connor PW, Sndberg-Wollheim M, Thompson AJ, Weinshenker BG, Wolinsky JS (2005) Dignostic criteri for multiple sclerosis: 2005 revisions to the McDonld Criteri. Ann Neurol 58: Chrlson ME, Pompei P, Ales KL, McKenzie CR (1987) A new method of clssifying prognostic comorbidity in longitudinl studies: development nd vlidtion. J Chronic Dis 40: Kingwell E, vn der Kop M, Zho Y, Shirni A, Zhu F, Oger J, Tremlett H (2012) Reltive mortlity nd survivl in multiple sclerosis: findings from British Columbi, Cnd. J Neurol Neurosurg Psychitry 83: Llmohmed A, Bzelier MT, Vn St TP, Uitdehg BM, Leufkens HG, De Boer A, De Vries F (2012) Cuses of deth in ptients with multiple sclerosis nd mtched referent subjects: popultion-bsed cohort study. Eur J Neurol 19: Lery E, Morrissey S, Younq J, Coustns M, Le Pge E, Chperon J, Edn G (2007) Long-term survivl of ptients with multiple sclerosis in West Frnce. Mult Scler 13: Sumelhti ML, Hkm M, Elovr I, Pukkl E (2010) Cuses of deth mong ptients with multiple sclerosis. Mult Scler 16: Bronnum-Hnsen H, Koch-Henriksen N, Stenger E (2004) Trends in survivl nd cuse of deth in Dnish ptients with multiple sclerosis. Brin 127: Grytten Torkildsen N, Lie SA, Arseth JH, Nylnd H, Myhr KM (2008) Survivl nd cuse of deth in multiple sclerosis: results from 50-yer follow-up in Western Norwy. Mult Scler 14: Rgonese P, Aridon P, Slemi G, D Amelio M, Svettieri G (2008) Mortlity in multiple sclerosis: review. Eur J Neurol 15: Pompili M, Forte A, Plermo M, Stefni H, Lmis DA, Serfini G, Amore M, Girrdi P (2012) Suicide risk in multiple sclerosis: systemtic review of current literture. J Psychosom Res 73: Pty DW, Li DK (1993) Interferon bet-1b is effective in relpsing-remitting multiple sclerosis. II. MRI nlysis results of multicenter, rndomized, double-blind, plcebo-controlled tril. UBC MS/MRI Study Group nd the IFNB Multiple Sclerosis Study Group. Neurology 43: The IFNB Multiple Sclerosis Study Group (1993) Interferon bet- 1b is effective in relpsing-remitting multiple sclerosis. I. Clinicl results of multicenter, rndomized, double-blind, plcebo-controlled tril. The IFNB Multiple Sclerosis Study Group. Neurology 43: Trojno M, Pellegrini F, Fuini A, Policelli D, Zipoli V, Zimtore GB, Di Monte E, Portccio E, Lepore V, Livre P, Amto MP (2007) New nturl history of interferon-bet-treted relpsing multiple sclerosis. Ann Neurol 61: Goodin DS, Ebers GC, Cutter G, Cook SD, O Donnell T, Reder AT, Kremenchutzky M, Oger J, Rmett M, Beckmnn K, Knppertz V (2012) Cuse of deth in MS: long-term follow-up of rndomised cohort, 21 yers fter the strt of the pivotl IFNbet-1b study. BMJ Open 2(6). doi: /bmjopen Goodin DS, Reder AT, Ebers GC, Cutter G, Kremenchutzky M, Oger J, Lngdon D, Rmett M, Beckmnn K, DeSimone TM, Knppertz V (2012) Survivl in MS: rndomized cohort study 21 yers fter the strt of the pivotl IFNbet-1b tril. Neurology 78: