Epidemiology of Insomni: Longitudinl Study in UK Popultion Hnnh Morphy, MMedSci 1, Kte M. Dunn, PhD 1 ; Mrtyn Lewis, PhD 1 ; Helen F. Bordmn, PhD 2 ; Peter R. Croft, MD 1 1 Primry Cre Musculoskeletl Reserch Centre, Keele University, Keele, UK; 2 Centre for Phrmcy, Helth nd Society, University of Nottinghm, Nottinghm, UK Study Objectives: To investigte the incidence, persistence, nd consequences of insomni nd their ssocitions with psychologicl helth nd pin. Design: A popultion bsed, longitudinl, cohort study using postl questionnires t bseline nd 12-month follow-up. Sleep problems in the pst month were ssessed using 4 questions: insomni ws defined s hving t lest 1 of the sleep problems on most nights. Questions bout psychologicl helth, presence of pin t different sites, nd demogrphic detils were included in the questionnire. Setting: Five generl prctices in Stffordshire, UK. Prticipnts: The questionnire ws miled to rndom smple of 4885 dults ged 18 yers nd over registered with these prctices. There were 2662 questionnires returned. Results: Of the responders, 2363 completed ll 4 sleep questions t bseline: 870 (37%) hd insomni nd 1493 (63%) did not hve insomni. Of those without insomni t bseline, the incidence of insomni t 12 INTRODUCTION Disclosure Sttement This ws not n industry supported study. Drs. Morphy, Dunn, Lewis, Bordmn, nd Croft hve reported no finncil conflicts of interest. Submitted for publiction My 11, 2006 Accepted for publiction December 24, 2006 Address correspondence to: Dr Hnnh Morphy, Primry Cre Musculoskeletl, Reserch Centre, Keele University, Stffordshire, ST5 5BG, UK; Tel: 44 0 1782 583905; Fx: 44 0 1782 583911; E-mil: h.morphy@cphc.keele. c.uk months ws 15%, nd this ws significntly ssocited with bseline nxiety, depression, nd pin. Of those who did hve insomni t bseline, 69% hd insomni t 12-month follow-up; persistence of insomni ws significntly ssocited with older ge. Insomni t bseline ws significntly ssocited with incidence of nxiety, depression, nd widespred pin t 12-month follow-up. Conclusions: Insomni is common nd often persistent. Older people pper more vulnerble to persistent symptoms. Our results provide evidence tht the common problems of insomni, pin, nd psychologicl distress re intertwined nd suggest tht combined pproches to tretment my be needed to reduce the onset nd persistence of these problems in the community. Keywords: Epidemiology, incidence, longitudinl studies, prevlence, sleep initition nd mintennce disorders Cittion: Morphy H; Dunn KM; Lewis M et l. Epidemiology of Insomni: Longitudinl Study in UK Popultion. SLEEP 2007;30(3):274-280. INSOMNIA IS COMMON AND INCLUDES SYMPTOMS SUCH AS DIFFICULTY FALLING ASLEEP OR STAYING ASLEEP, EARLY MORNING WAKENING, AND SLEEP disstisfction, s well s dytime consequences such s tiredness. 1 Studies of generl-popultion smples hve found the prevlence of insomni to rnge between 10% nd 48%. 2,3 Much of this vrition in prevlence my be explined by the use of different definitions of insomni in these studies. 1 Insomni cn hve importnt consequences for sufferers nd my impct on helth, work, nd qulity of life. Such ssocitions include the development of depression, 2 incresed use of helth cre services 4,5 nd higher risk of motor vehicle ccidents. 6 In ddition, studies looking t the reltionship between insomni nd work hve found ssocitions with bsenteeism, decresed concentrtion, impired work performnce nd work relted ccidents. 5,7 Cross-sectionl studies of insomni hve consistently found femle sex, 1,2,8 mentl helth problems, 2,4,9,10 nd physicl helth problems 11-14 to be ssocited with n incresed prevlence. A link between incresing ge nd insomni hs been reported, 1,4 lthough this finding is not consistent. 8,15,16 A smll number of epidemiologic studies hve looked t the reltionship between pin nd insomni. 15,17 Although there re numerous cross-sectionl studies investigting prevlence of insomni in the USA 11,18-20 nd Europe, 3,4,16,21-25 including the UK, 9 less is known bout the onset nd nturl history of insomni. Mny of these studies hve highlighted fctors ssocited with prevlent insomni, nd it is not cler wht is cuse nd effect or whether the fctors re linked with the onset or persistence of the problem. Longitudinl studies cn help to unrvel these ssocitions. The ims of this study were to investigte the incidence nd persistence of insomni nd fctors ssocited with these, nd insomni s risk fctor for nxiety, depression, nd pin, in longitudinl study of n dult, UK, generl-popultion smple. MATERIALS AND METHODS Study Popultion nd Design A postl survey of popultion ged 18 yers nd over ws crried out using self-completion questionnire. The primry focus of the study ws on hedche, but the survey contined wide rnge of questions, including items ddressing sleep problems. 26 The dult popultion registered with 5 generl prctices in the UK, covering mix of urbn nd rurl res, provided the smpling frme. One thousnd ptients were rndomly selected from ech prctice register. More thn 95% of people in the UK re registered with generl prctitioner (regrdless of whether they consult or not), nd prctice registers provide convenient frme for smpling the locl popultion. Doctors from the prctices checked the selected ptients for exclusions. One hundred nd fifteen ptients were excluded, the mjority becuse they hd died or left the prctice (becuse they moved) in the 6 weeks between obtining the smple nd conducting the miling, nd smll proportion becuse they were in the hospitl t the time SLEEP, Vol. 30, No. 3, 2007 274
of the survey or were suffering from serious mentl illness. This ment tht 4885 questionnires were sent out t bseline. Ethicl pprovl for the study ws obtined from the Locl Reserch Ethics Committee. Dt were collected using self-completion questionnires, miled with study informtion leflet, letter from the generl prctice, nd form requesting consent to further involvement in the study. Reminders were sent to nonrespondents t 2 nd 5 weeks. 26 The bseline questionnire ws miled in April 2000, nd follow-up questionnire (repeting the sme questions) ws sent 1 yer lter to those people who responded to the first questionnire nd who consented to further involvement. 27 Further detils of the survey hve been reported elsewhere. 26-28 Questionnire The questionnire included sections on generl helth, demogrphics, nd hedche. It included the Hospitl Anxiety nd Depression Scle s mesure of nxiety nd depression 29 ; this instrument hs been vlidted nd shown to be useful in identifying depression nd nxiety in generl-popultion smples. 30,31 The Hospitl Anxiety nd Depression Scle ws scored ccording to pressigned clssifiction of noncse (score 0-7), possible cse (score 8-10), nd probble cse (score 11 nd over) for the seprte dimensions of nxiety nd depression, s specified by the uthors. 29 A blnk body mnikin (with front nd bck views) sking prticipnts to shde res of ny ches or pins lsting for 1 dy or more in the previous month ws used to identify number of pin res (rnging from 0 to 7). This ws clssified ccording to criteri of Mcfrlne 32 with the ddition of the hed re. Due to the underlying ordinl nture of the scle nd the fct tht subgroup nlyses will result in smll frequency counts in certin cells, number of pin re ctegories were grouped s 0, 1, 2 to 3, nd 4 to 7. The relibility of questions on hedche, bodily pin, nxiety, depression, nd sleep problem ws exmined s prt of the wider study; they were found to be relible for use in the generl popultion. 33 Questions bout employment sttus nd occuption (own nd prtner) were used to determine socil clss. Socil clss ws grouped s I professionl, II mngeril nd technicl, IIIN nonmnul skilled, IIIM mnul skilled, IV prtly skilled, nd V unskilled. Groups I to IIIN were then clssed s nonmnul nd IIIM to V s mnul. 34,35 There were 4 questions relting to sleep, bsed on work by Jenkins et l. 36 These were: Over the lst month did you: () hve trouble flling sleep, (b) wke up severl times per night, (c) hve trouble stying sleep, nd (d) wke up fter your usul mount of sleep feeling tired nd worn out? There were 3 possible responses to ech question not t ll, on some nights, nd on most nights. Our min definition of insomni ws bsed on respondents nswering 1 or more questions on most nights. Since expert opinion differs regrding the clssifiction of insomni, prticulrly relting to the issue of nonrestortive sleep, we lso crried out sensitivity nlysis of our dt in reltion to 2 other plusible definitions of insomni. The first reclssifiction ws bsed on hving problems on most nights ccording to 1 or more of the nighttime symptoms only (trouble flling sleep nd/ or wke up severl times per night nd/or trouble stying sleep), ie, excluding the dytime symptom (wke up feeling tired nd worn out). The second reclssifiction defined insomni s endorsing 1 or more of the 3 nighttime symptoms on most nights nd the dytime symptom on most nights. Anlysis Three seprte nlyticl pproches were crried out to investigte the ssocitions with insomni. The first nlysis (incidence nlysis) estimted the incidence nd risk fctors for the development of insomni nd included prticipnts who responded to both the bseline nd 1-yer follow-up questionnires nd who hd no insomni t bseline. The second nlysis (persistence nlysis) investigted the frequency of persistent insomni t follow-up nd the fctors predicting the persistence of insomni. This nlysis included respondents who reported hving insomni t bseline nd who completed the 1-yer follow-up. For the incidence nd persistence nlyses, the ssocitions between insomni t 1-yer follow-up nd demogrphic nd generl helth vribles t bseline were exmined. The third nlysis (consequence nlysis) investigted whether the presence of insomni t bseline ws risk fctor for developing nxiety, depression, or pin t follow-up. This involved 3 subgroup nlyses: the first bsed on respondents with no nxiety ( noncse or possible cse ) t bseline, with nxiety ( probble cse) or no nxiety t 12-month follow-up s the outcome; the second on respondents with no depression ( noncse or possible cse ) t bseline, with depression ( probble cse) or no depression t 12-month follow-up s the outcome; the third on respondents with no widespred pin (defined s fewer thn 4 regions of pin) t bseline, with widespred pin (4 or more regions of pin) or no widespred pin t 12-month follow-up s the outcome. Associtions were mesured by risk rtios (RRs). Firstly, univrite nlyses were crried out to investigte the crude ssocitions between outcomes nd individul bseline risk fctors. Secondly, multivrite nlyses were crried out to evlute the independent effects on outcome of risk fctors djusted for covrites. The multivrite estimtion ws crried out using Cox regression (pplying constnt risk period 37 ). Associtions with ge re shown in terms of RRs per incrementl unit chnge of 10 yers. Sttisticl significnce is given t the 5% probbility level (2 tils). All nlyses were crried out using SPSS 12.0 (SPSS, Inc., Chicgo, IL) nd Confidence Intervl Anlysis (version 2.0.0. University of Southmpton School of Medicine, Southmpton, UK). RESULTS Response A totl of 2662 completed questionnires were returned t bseline, giving n djusted response of 56% following djustment for deths nd incorrect ddresses occurring or scertined fter miling. Response ws higher in women thn men (61% versus 51%, respectively) nd higher in older thn younger groups (65% in the over 65 yers ge group compred to 43% in the 18- to 35-yer ge group). Only questionnires in which ll the questions relted to sleep were nswered were used in the nlysis. There were 299 questionnires with missing dt on 1 or more of the sleep questions, mening tht 2363 respondents were included in the cohort for this study. The medin ge of the smple ws 50 yers (rnge 18 to 98 yers), nd 55% of the smple were women. Of the respondents to the bseline survey, 84% consented to receiving nother questionnire. Eighty-four people hd left the SLEEP, Vol. 30, No. 3, 2007 275
Tble 1 Chrcteristics of Responders t Bseline nd 12-Month Follow-up Bseline Follow-up (n = 2662) (n = 1589) Age, y 51.7 (17.5) 52.3 (16.3) Sex Men 1175 (44.1) 697 (43.9) Women 1487 (55.9) 892 (56.1) Socil clss Nonmnul 1451 (59.9) 953 (63.9) Mnul 972 (40.1) 539 (36.1) Anxiety Noncse 1525 (60.4) 921 (61.2) Possible cse 515 (20.4) 289 (19.2) Probble cse 483 (19.1) 295 (19.6) Depression Noncse 1984 (79.4) 1225 (80.2) Possible cse 316 (12.6) 185 (12.1) Probble cse 200 (8.0) 117 (7.7) Pin res 0 758 (28.5) 162 (10.2) 1 348 (13.1) 440 (27.7) 2-3 693 (26.0) 392 (24.7) 4-7 863 (32.4) 595 (37.4) Sleep questions Trouble flling sleep None 953 (37.8) 582 (38.3) Some nights 1259 (50.0) 758 (49.9) Most nights 306 (12.2) 180 (11.8) Wke up severl times t night None 575 (23.0) 306 (20.4) Some nights 1249 (50.2) 795 (52.9) Most nights 664 (26.7) 401 (26.7) Trouble stying sleep None 897 (40.0) 480 (32.8) Some nights 1101 (45.4) 729 (49.9) Most nights 427 (17.6) 253 (17.3) Wke up tired nd worn out None 696 (30.0) 428 (28.6) Some nights 1307 (52.4) 803 (53.7) Most nights 491 (19.7) 265 (17.7) Insomni (min definition) No 1493 (63.2) 933 (65.2) Yes 870 (36.8) 498 (34.8) Dt re presented s numbers (percentges), except ge, which is men ± SD. Numbers do not lwys dd to the mrginl totls (2662 t bseline; 1589 t follow up) due to some missing dt. Trouble flling sleep (most nights) nd/or Wke up severl times t night (most nights) nd/or Hve trouble flling sleep (most nights) nd/or Wke up tired nd worn out (most nights). prctice or died during follow-up; therefore 2141 questionnires were miled t 1 yer nd 1589 were returned, giving response to follow-up of 74%. There ws little difference in response to follow-up between those with nd without insomni t bseline. Chrcteristics of responders to the bseline nd 12-month questionnires re presented in Tble 1. The prevlence of insomni t bseline ws 36.8% (95% confidence intervl 34.9, 38.8). Tble 1 summrizes the individul responses to the 4 sleep questions. The lest common symptom t bseline ws trouble stying sleep, to which 40% nswered not t ll. The most common symptom ws wke up severl times t night, which SLEEP, Vol. 30, No. 3, 2007 276 ffected 27% of respondents on most nights t bseline nd 12 months. Incidence Anlysis Among the 859 respondents who were included in the incidence nlysis (ie, those with no insomni t bseline who responded to the sleep questions t follow-up), 125 were clssified with insomni t 1 yer, n incidence of 14.6% (95% confidence intervl 12.2, 16.9). Tble 2 shows the ssocitions between bseline demogrphic nd generl helth fctors nd the incidence of insomni t 12 months. Age, sex, nd socil sttus were not significntly ssocited with incidence of insomni. However, depression nd nxiety were ssocited with significnt increse in the development of insomni in both undjusted nd multivrite nlyses. Incidence of insomni lso incresed with the number of pin res reported t bseline. Depression ws the strongest independent predictor of developing insomni, with probble cses of depression hving 3 times the risk of developing insomni compred with noncses. Persistence Anlysis The persistence nlysis included 483 respondents: those who were clssified s hving insomni t bseline who lso responded to the sleep questions t 12 months. Of these, 334 continued to report insomni t follow-up, giving persistence estimte of 69.2% (95% confidence intervl 65.0, 73.3). Tble 3 shows the ssocitions between bseline demogrphic nd generl helth fctors nd the persistence of insomni t 12 months. The only bseline fctor significntly relted to the persistence of insomni ws ge. After djusting for the other vribles, we estimted tht ech 10-yer increse in ge ws ssocited with n elevted risk of 1.1 of hving persistent insomni. Given tht persistence ffected bout 70% nd tht the men ge of our study popultion ws pproximtely 50, our estimted RR of 1.1 would imply tht someone ged 60 with bseline insomni ws likely to hve 7% greter bsolute risk (1.1 70%) of hving persistent pin t follow-up, compred with someone ged 50; the bsolute risk difference is pproximtely 15% (1.1 1.1 70%) for 70- compred with 50-yer-olds. Consequences Anlysis Results relting to the 3 consequences nlyses re shown in Tble 4. Insomni t bseline ws significntly ssocited with incidence of nxiety, depression, nd widespred pin t 12 months. The ssocitions were sttisticlly significnt both before nd fter djustment for bseline covrites. After djustment, insomni t bseline ws ssocited with nerly 3 times the risk of developing depression nd more thn 2 times the risk of developing nxiety t 12-month follow-up. Sensitivity Anlysis Ptterns of ssocition, when the 2 lterntive definitions of insomni were pplied, showed brod similrities with the min nlyses bove, lthough some becme weker or lost sttisticl significnce. The prevlence of 1 or more sleep problems on most nights cross the 3 nighttime symptoms ws 30.4% (719/2363); incidence ws 13.3% (126/949), nd persistence 67.9% (267/393).
Tble 2 Incidence of Insomni t 12-Month Follow-up Insomni No insomni Undjusted RR Adjusted RR b No. (%) No. (%) (95% CI) (95% CI) Subgroup totls c 125 (14.6) 734 (85.4) - - Age - - 1.06 (0.95, 1.20) d 1.06 (0.94, 1.20) d Sex Men e 53 (12.8) 360 (87.2) 1.00 1.00 Women 72 (16.1) 374 (83.9) 1.26 (0.88, 1.79) 1.25 (0.85, 1.82) Socil clss Nonmnul e 80 (14.3) 481 (85.7) 1.00 1.00 Mnul 41 (15.7) 220 (84.3) 1.10 (0.76, 1.61) 1.15 (0.78, 1.68) Anxiety Noncse e 64 (10.2) 557 (89.7) 1.00 1.00 Possible cse 26 (20.0) 104 (80.0) 2.42 (1.61, 3.64) 2.13 (1.36, 3.31) Probble cse 33 (38.8) 52 (61.2) 3.16 (1.98, 5.05) 2.43 (1.39, 4.24) Depression Noncse e 88 (11.4) 682 (88.6) 1.00 1.00 Possible cse 22 (41.5) 31 (58.5) 1.73 (0.99, 3.02) 1.04 (0.55, 1.97) Probble cse 13 (56.5) 10 (43.5) 5.74 (2.90, 11.4) 3.21 (1.49, 6.91) Pin res 0 e 25 (8.9) 255 (91.1) 1.00 1.00 1 17 (12.2) 122 (87.8) 1.37 (0.74, 2.54) 1.30 (0.70, 2.42) 2-3 38 (16.3) 195 (83.7) 1.83 (1.10, 3.03) 1.67 (0.99, 2.80) 4-7 45 (21.7) 162 (78.3) 2.44 (1.49, 3.97) 1.71 (1.02, 2.87) CI refers to confidence intervl. Insomni defined s: Trouble flling sleep (most nights) nd/or Wke up severl times t night (most nights) nd/or Hve trouble flling sleep (most nights) nd/or Wke up tired nd worn out (most nights). b Risk rtios (RR) djusted for ge, sex, socil clss, nxiety, depression, pin res. c Subgroup totls denote the number of respondents with insomni nd the number of respondents without insomni t follow-up mong those who did not hve insomni t bseline. Numbers my not dd up to the subgroup totls due to some missing dt. d Risk rtios relte to unit increses in ge of 10 yers. e Reference ctegory for clcultion of risk rtios. After djustment, incident insomni ws significntly ssocited with incresed ge (RR = 1.18); femle sex (RR = 1.50); possible nxiety (RR = 2.17); probble nxiety (RR = 2.09); probble depression (RR = 2.77); 2 to 3 pin res (RR = 1.97), nd 4 to 7 pin res (RR = 2.01). After djustment, persistent insomni ws significntly ssocited with incresed ge (RR = 1.10). Insomni ws significntly ssocited with incident depression (RR = 2.19) nd widespred pin (RR = 1.56) nd linked with incident nxiety but not significntly (RR = 1.54). The prevlence for the stricter definition of ny nighttime symptom on most nights plus the dytime symptom of feeling tired nd worn out on most nights ws 13.2% (313/2363); incidence ws 6.8% (80/1176), nd persistence 54.8% (91/166). After djustment, incident insomni ws significntly ssocited with femle sex (RR = 1.98); probble nxiety (RR = 2.31); probble depression (RR = 4.58), nd 4 to 7 pin res (RR = 2.24). After djustment, persistent insomni ws not significntly ssocited with ny fctor, though ge hd similr point estimte s in the min nlysis (RR = 1.10). Insomni ws significntly ssocited with incident depression (RR = 2.57) but showed little ssocition with incident nxiety (RR = 1.17) nd widespred pin (RR = 1.18). DISCUSSION Our findings confirm tht insomni is common in the dult UK generl popultion, with prevlence of 37%. We hve shown tht much of this insomni is persistent, with over two thirds (69%) of sufferers t bseline lso reporting symptoms yer lter. The incidence of insomni is 15% in this popultion nd is ssocited with preexisting nxiety, depression, nd pin s reported t bseline. In contrst, the persistence of insomni ws significntly ssocited only with older ge. We hve lso provided evidence tht insomni is risk fctor for the subsequent onset of nxiety, depression, nd pin. A mjor strength of this study is the longitudinl design, enbling us to investigte nd compre the fctors ssocited with incidence, persistence, nd consequences of insomni over the course of 1 yer. Numerous studies hve looked t the prevlence nd cross-sectionl ssocitions of insomni. However, such studies re not ble to seprte fctors tht influence the development or persistence of insomni or to investigte insomni s risk fctor for other conditions. Although there re longitudinl studies investigting insomni, 2,13,38 none of the studies identified hd compred fctors relted to the incidence of insomni with those relted to persistence in generl dult popultion. The popultion bsis (rurl nd urbn res) nd lrge smple size re lso importnt strengths of this study nd enhnce the generlizbility of the findings. In ddition, fctors such s pin nd psychologicl sttus were mesured using instruments vlidted for use in the generl popultion. Compring occurrence figures between studies is difficult. The min resons relte to differences in definitions of insomni, differences in survey methodologies, nd lso the vrition in length SLEEP, Vol. 30, No. 3, 2007 277
Tble 3 Persistence of Insomni t 12-Month Follow-up Insomni No insomni Undjusted RR Adjusted RR b No. (%) No. (%) (95% CI) (95% CI) Subgroup totls c 334 (69.2) 149 (30.8) - - Age - - 1.09 (1.03, 1.17) d 1.10 (1.03, 1.18) d Sex Men d 117 (65.0) 63 (35.0) 1.00 1.00 Women 217 (71.6) 86 (28.4) 1.10 (0.88, 1.38) 1.13 (0.89, 1.43) Socil clss Nonmnul e 193 (67.2) 94 (32.8) 1.00 1.00 Mnul 129 (71.7) 51 (28.3) 1.07 (0.85, 1.33) 1.06 (0.84, 1.33) Anxiety Noncse e 116 (65.9) 60 (34.1) 1.00 1.00 Possible cse 94 (69.6) 41 (30.4) 1.06 (0.81, 1.39) 1.05 (0.79, 1.40) Probble cse 120 (74.1) 42 (25.9) 1.12 (0.87, 1.45) 1.12 (0.81, 1.53) Depression Noncse e 193 (64.3) 107 (35.7) 1.00 1.00 Possible cse 79 (77.5) 23 (22.5) 1.20 (0.93, 1.56) 1.11 (0.82, 1.48) Probble cse 55 (77.5) 16 (22.5) 1.20 (0.89, 1.63) 1.08 (0.75, 1.54) Pin res 0 e 52 (64.2) 29 (35.8) 1.00 1.00 1 24 (66.7) 12 (33.3) 1.04 (0.64, 1.68) 1.02 (0.62, 1.68) 2-3 74 (66.1) 38 (33.9) 1.03 (0.72, 1.47) 1.04 (0.71, 1.51) 4-7 184 (72.4) 70 (27.6) 1.13 (0.83, 1.54) 1.09 (0.78, 1.51) Insomni defined s Trouble flling sleep (most nights) nd/or Wke up severl times t night (most nights) nd/or Hve trouble flling sleep (most nights) nd/or Wke up tired nd worn out (most nights). b Risk rtios djusted for ge, sex, socil clss, nxiety, depression, pin res. c Subgroup totls denote the number of respondents with insomni nd the number of respondents without insomni t follow-up mong those who did hve insomni t bseline. Numbers my not dd up to the subgroup totls due to some missing dt. d Risk rtios relte to unit increses in ge of 10 yers. e Reference ctegory for clcultion of risk rtios. Tble 4 Consequences of Insomni t 12-Month Follow-up Problem t follow up Insomni b Yes No Undjusted RR Adjusted RR c t bseline No. (%) No. (%) (95% CI) (95% CI) Anxiety No d 47 (6.0%) 737 (94.0%) 1.00 1.00 Yes 54 (16.7%) 270 (83.4%) 2.78 (1.88, 4.11) 2.28 (1.43, 3.64) Depression No d 17 (2.0%) 850 (98.0%) 1.00 1.00 Yes 35 (8.4%) 384 (91.7%) 4.26 (2.39, 7.60) 2.71 (1.37, 5.37) Widespred pin No d 127 (18.2%) 569 (81.8%) 1.00 1.00 Yes 71 (28.3%) 180 (71.7%) 1.55 (1.16, 2.07) 1.45 (1.03, 2.03) Problem t follow-up refers to incident nxiety, depression or widespred pin. Anxiety nd depression were bsed on the clssifiction of probble cse (no nxiety nd no depression being bsed on noncse nd possible cse criteri). Widespred pin ws denoted by pin in four or more regions of the mnikin (no widespred pin is denoted by none to 3 res of pin). In ech of the 3 nlyses the denomintor figures re those with no problem t bseline. b Insomni defined s Trouble flling sleep (most nights) nd/or Wke up severl times t night (most nights) nd/or Hve trouble flling sleep (most nights) nd/or Wke up tired nd worn out (most nights). c Risk rtios djusted for ge, sex, socil clss, nxiety (except when nxiety is the problem of interest), depression (except when depression is the problem of interest), pin res (except when widespred pin is the problem of interest). d Reference ctegory for clcultion of risk rtios. of follow-up in longitudinl studies. Studies hve reported similr prevlence estimtes to our min definition. 4,8,11 However, studies with stricter criteri for defining insomni tend to report lower prevlence nd incidence figures. For exmple, study by Ford nd Kmrow, 2 which used stricter definition thn our min definition, reported prevlence of 10.0% nd n incidence of SLEEP, Vol. 30, No. 3, 2007 278 6.2% (compred with 36.8% nd 14.6% in our study). Similrly, our conservtive definition (requiring both nighttime nd dytime symptom) gve lower prevlence nd incidence estimtes of 13.2% nd 6.8%, respectively. There is some debte s to whether the symptom wke up tired nd worn out my be symptom of other conditions such s sleep pne rther thn being truly
insomni. 39 This item ws included within our principl definition of insomni, s recommended by the uthors of the questionnire tht we employed. Excluding this dytime symptom, while mintining the criterion of sleep problems on most nights for t lest 1 of the 3 nighttime symptoms, the prevlence, incidence, nd persistence estimtes were little chnged: 30.4%, 13.3%, nd 67.9%, respectively. Furthermore, the different definitions of insomni did not gretly influence the sttisticl ssocitions. The min difference occurred when the dytime symptom wke up feeling tired nd worn out on most nights ws mde mndtory to the clssifiction of insomni, resulting in reduction in the risk of developing nxiety nd pin mong persons with insomni. Older people were more likely thn younger people to hve persistent insomni. There is complex ssocition between ge nd insomni. Some studies hve found cler reltionship between incresing ge nd insomni, 4,11,23 but other studies hve found no ssocition. 9,16 Plleson et l 24 found tht erly morning wkening nd difficulty mintining sleep were more common in the older ge group, wheres dytime impirment ws less common in this ge group. However, ssocitions between incresing ge nd the persistence of insomni were consistent cross the 3 seprte definitions of insomni considered in our study. There hve only been smll number of studies investigting the reltionship between pin nd insomni in generl popultion smples. A study by Ohyon 17 looked t the ssocition between chronic pin (6 months durtion or longer) t different sites nd insomni nd found strong reltionship between insomni nd chronic pin. A study from Cnd lso found tht pin ws significntly ssocited with insomni. 15 Our study found tht reporting of 4 or more pin res ws significnt risk fctor for the subsequent incidence of insomni independent of depression nd nxiety. There ws higher incidence of insomni s the number of reported pin res incresed. Also, insomni ws shown to be risk fctor for the subsequent development of pin. Psychologicl helth problems, such s nxiety nd depression, hve been consistently linked with insomni in number of popultion studies. 2,4,13,20 The findings of Foley et l, 13 bsed on 3- yer longitudinl study of 6800 dults ged 65 yers nd older reported tht depression ws linked with the incidence nd retention of insomni. In our study, both nxiety nd depression t bseline were significntly linked with incidence, but not persistence, of insomni. Moreover, our study lso provides evidence indicting tht insomni is significnt risk fctor for the subsequent onset of nxiety nd depression. One potentil limittion of this study is the bseline response of 56%, giving potentil for bis. In order to investigte this, n ge-sex stndrdized prevlence figure ws clculted nd ws very similr to our crude estimte (stndrdized estimte 36.4%, crude estimte 36.8%). Furthermore, there ws little evidence of response bis when compring those who replied erly to the questionnire nd those who replied only fter reminder. Although unknown bises in bseline response my ffect the ccurcy of the prevlence figure, the estimtes of ssocitions mong other bseline fctors nd prevlence, incidence, nd persistence of insomni re unlikely to be ffected. In terms of bis in followup, there were few bseline differences between respondents nd nonrespondents to the 12-month questionnire. A second potentil limittion is tht our prevlence figure reltes to 1 popultion district in the UK. This re is socioeconomiclly more deprived thn verge for the UK nd hs reltively smll ethnic minority popultion. However, these differences re not substntil nd re unlikely to gretly ffect the generlizbility of the findings, prticulrly the prospective ssocitions. Prevlence is useful in highlighting the size of problem, nd insomni is clerly common. Although not ll persons reporting insomni will regrd it s problem, there is evidence from fmily prctice settings tht high proportion of consulters hve insomni symptoms. 40 Further reserch would be needed to clrify the benefits of identifying nd deling with the problem, but the strong links mong insomni nd nxiety, depression, nd pin suggest its possible involvement in rnge of presenting problems. Cross-sectionl prevlence studies cnnot disentngle cuse nd effect, but it is likely tht some cses of insomni result from other longstnding conditions. Our results confirm this for depression in prticulr, but nxiety nd pin were lso significnt predictors of incident insomni. Evlution of the popultion ttributble proportions bsed on our djusted estimtes of RR, estimte tht 21% of incident insomni is sttisticlly linked with prior nxiety, 19% with the prior presence of 4 or more res of pin, nd 10% with depression; the percentge ttributble to depression is less (despite stronger ssocition) becuse there were reltively few respondents with probble depression. However the reltionships re 2 wy, given our findings tht insomni predisposes to the onset of depression, nxiety, nd pin, nd this indictes complex cycles of cuse nd effect in the development of these chronic problems. Our study suggests tht effective tretment of the individul problems when they first present hs the potentil to substntilly reduce the onset nd persistence of multiple morbidity in the generl popultion. This represents chllenge for public helth nd primry cre to identify strtegies for both tretment nd prevention of insomni. With incresing reluctnce to use sedtive mediction, self-help nd psychologicl pproches offer lterntive models of cre. Interventions trgeted t persons t higher risk my help reduce long-term problems, prticulrly mong older people. ACKNOWLEDGMENTS The Proprietry Assocition of Gret Britin funded the bseline survey. Other study costs were funded by Primry Cre Sciences Reserch Centre, the Deprtment of Medicines Mngement, Keele University nd the North Stffordshire Primry Cre Consortium. REFERENCES 1. Ohyon MM. Epidemiology of insomni: wht we know nd wht we still need to lern. Sleep Med Rev 2002;6:97-111. 2. Ford DE, Kmerow DB. Epidemiologic study of sleep disturbnces nd psychitric disorders. An opportunity for prevention? JAMA 1989;262:1479-84. 3. Quer-slv MA, Orluc A, Goldenberg F, Guilleminult C. Insomni nd use of hypnotics: study of French popultion. Sleep 1991;14:386-91. 4. Weyerer S, Dilling H. Prevlence nd tretment of insomni in the community: Results from the Upper Bvrin field study. Sleep 1991;14:392-8. 5. Leger D, Guilleminult C, Bder G, Lévy E, Pillrd M. Medicl nd socio-professionl impct of insomni. Sleep 2002;25:625-9. 6. Ohyon MM, Smirne S. Prevlence nd consequences of insomni disorders in the generl popultion of Itly. Sleep Med 2002;3:115- SLEEP, Vol. 30, No. 3, 2007 279
20. 7. Linton SJ, Bryngelsson I-L. Insomni nd its reltionship to work nd helth in working-ge popultion. J Occup Rehbil 2000;10:169-83. 8. Leger D, Guilleminult C, Dreyfus JP, Delhye C, Pillrd M. Prevlence of insomni in survey of 12,778 dults in Frnce. J Sleep Res 2000;9:35-42. 9. Ohyon MM, Culet M, Priest RG, Guilleminult C. DSM-1V nd ICSD-90 insomni symptoms nd sleep disstisfction. Br J Psychitry 1997;171:382-8. 10. Tylor DJ, Lichstein KL, Durrence HH et l. Epidemiology of insomni, depression, nd nxiety. Sleep 2005;28:1457-64. 11. Klink M, Qun SF. Prevlence of reported sleep disturbnces in generl dult popultion nd their reltionship to obstructive irwys diseses. Chest 1987;91:540-6. 12. Novk M, Mucsi I, Shpiro CM, Rethelyi J, Kopp MS. Incresed utiliztion of helth services by insomnics-n epidemiologicl perspective. J Psychosom Res 2004;56:527-36. 13. Foley DJ, Monjn A, Simonsick EM, Wllce RB, Blzer DG. Incidence nd remission of insomni mong elderly dults: n epidemiologicl study of 6,800 persons over three yers. Sleep 1999;22:366-72. 14. Tylor DJ, Lichstein KL, Durrence HH et l. Insomni s helth risk fctor. Behv Sleep Med 2003;1:227-47. 15. Sutton DA, Moldofsky H, Bdley EM. Insomni nd helth problems in Cndins. Sleep 2001;24:655-70. 16. Liljenberg B, Almqvist M, Hett J, Roos B-E, Ågren H. The prevlence of insomni: the importnce of opertionlly defined criteri. Ann Clin Res 1988;20:393-8. 17. Ohyon MM. Reltionship between chronic pinful condition nd insomni. J Psychitric Res 2005;39:151-9. 18. Ancoli-Isrel S, Roth T. Chrcteristics of insomni in the United Sttes: results of the 1991 Ntionl Sleep Foundtion survey. Sleep 1999;22:347-53. 19. Krcn I, Thornby JI, Anch CE et l. Prevlence of sleep disturbnce in primrily urbn Florid county. Soc Sci Med 1976;10:239-44. 20. Mellinger GD, Blter MB, Uhlenhuth EH. Insomni nd its tretment. Arch Gen Psychitry 1985;42:225-32. 21. Jnson C. Gislson T, De Bcker W et l. Prevlence of sleep disturbnces mong young dults in three Europen countries. Sleep 1995;18:589-97. 22. Lugresi E, Zucconi M, Bixler E. Epidemiology of sleep disorders. Psychitr Ann 1987;17:446-53. 23. Ohyon MM. Epidemiologicl study on insomni in the generl popultion. Sleep 1996;19:7-15. 24. Plleson S, Nordhus IH, Nielsen GH, et l. Prevlence of insomni in the dult Norwegin popultion. Sleep 2001;24:771-9. 25. Tribl GG, Schmeiser-Reider A, Rosenberger A, et l. Sleeping hbits in the Austrin popultion. Sleep Med 2002;3:21-8. 26. Bordmn HF, Thoms E, Croft PR, Millson DS. Epidemiology of hedche in n English district. Cephlgi 2003;23:129-37. 27. Bordmn HF, Thoms E, Millson DS, Croft PR. One-yer follow-up of hedche in n dult generl popultion. Hedche 2005;45:337-45. 28. Bordmn HF, Thoms E, Millson DS, Croft PR. Psychologicl, sleep, lifestyle, nd comorbid ssocitions with hedche. Hedche 2005;45:657-69. 29. Zigmond AS, Snith RP. The hospitl nxiety nd depression scle. Act Psychitr Scnd 1983;67:361-70. 30. Bjellnd I, Dhl AA, Hug TT, Neckelmnn D. The vlidity of the Hospitl Anxiety nd Depression Scle. An updted literture review. J Psychosom Res 2002;52:69-77. 31. Lisspers J, Nygren A, Sodermn E. Hospitl Anxiety nd Depression Scle (HAD): some psychometric dt for Swedish smple. Act Psychitr Scnd 1997;96:281-6. 32. Mcfrlne GJ. Generlized pin, fibromylgi nd regionl pin: SLEEP, Vol. 30, No. 3, 2007 280 epidemiologicl view. Blliere s Clin Rheumtol 1999;13:403-14. 33. Bordmn HF, Thoms E, Millson DS, et l. North Stffordshire Hedche Survey: development, relibility nd vlidity of questionnire for use in generl popultion survey. Cephlgi 2003;23:325-31. 34. Croft PR, Lewis M, Ppgeorgiou AC, et l. Risk fctors for neck pin: longitudinl study in the generl popultion. Pin 2001;93:317-25 35. Stndrd occuptionl clssifiction, Vol. 3. Socil clssifictions nd coding methodology. London: HMSO; 1991. 36. Jenkins CD, Stnton B-A, Niemcryk SJ, Rose RM. A scle for the estimtion of sleep problems in clinicl reserch. J Clin Epidemiol 1988;41:313-21. 37. Thompson ML, Myers JE, Kriebel D. Prevlence odds rtio or prevlence rtio in the nlysis of cross sectionl dt: wht is to be done? Occup Environ Med 1998;55:272-7. 38. Mllon L, Bromn J-E, Hett J. Reltionship between insomni, depression, nd mortlity: 12-yer follow-up of older dults in the community. Int Psychogeritr 2000;12:295-306. 39. Lichstein KL, Riedel BW, Lester KW, Aguillrd RN. Occult sleep pne in recruited smple of older dults with insomni. J Consult Clin Psychol 1999;67:405-10. 40. Hoygen F, Rink K, Käppler C et l. Prevlence nd tretment of insomni in generl prctice. A longitudinl study. Eur Arch Psychitry Clin Neurosci 1993;242:329-36.