Prevalence of sleep disturbance in a primarily urban Florida county

Prevalence of sleep disturbance in a primarily urban Florida county

Sot SCI& Mud,Vol IO pp 239to 244 Pergcsmon Press1976Prmtedm GreatBnum PREVALENCE OF SLEEP DISTURBANCE IN A PRIMARILY URBAN FLORIDA COUNTY* ISMET CHAR...

738KB Sizes 13 Downloads 39 Views

Sot SCI& Mud,Vol IO pp 239to 244 Pergcsmon Press1976Prmtedm GreatBnum

PREVALENCE OF SLEEP DISTURBANCE IN A PRIMARILY URBAN FLORIDA COUNTY* ISMET CHARLES

KARACAN, JOHN I THORNBY, MICHAEL ANCH,

E. HOLZER~.,GEORGE J WARHEIT~,JOHN J SCHWABQand ROBERT L. WILLIAMS

Department

of Psychiatry, Baylor College of Medtcme and Veterans Admmrstratton m Houston, Texas

Hospttal

Abstract-Data for thts paper ortgtnated from the Flortda Health Study Program, a comprehenstve survey ehcttmg general health mformatton from a random sample of 1645 mdtvrduals m Alachua County, Florida Patterns of soao-demographic differences were studied for each of SIX questtonnatre Items relating to sleep More than one-thud of respondents reported having trouble with sleep This was greatest for older non-single females at a low SES level but unrelated to race Use of sleeping medtcattons was highest among older white females, but unrelated to race, SES, and marital status Life-style tmphcattons were reflected m types of sleeping dtfhculty Important contrrbuttons to our understandmg of the scientific measurement of sleep have been achieved through polygraphic momtormg of the electroencephalogram (EEG) However, this method of sleep analysts IS expensive, time-consuming and necessartly ltmtted m tts capacity to generate epldemioloqcal type data for a large population of mdivtduals

The sample survey 1s an approach to understandmg the prevalence of sleep problems m the community havmg many obvious advantages over the modem laboratory assessment One maJor advantage is the opportunity to select a large representative sampling of the population and therefore avoid the potential biases m selection associated with the much smaller number of individuals who can be attracted to the sleep laboratory Furthermore, the survey is an efi% cient way of obtammg a broad spectrum of mformation for the populatton of mdividuals suffermg from various types of sleep disorders The survey should be considered as a valuable adJunct to laboratory momtormg It alone can provtde subJective mformation concerning the dtstribution and severity of sleep disorders m the commumty, thereby revealmg evtdence of problems most common to the population These can then be studied m greater depth usmg electroencephalographic momtormg Effects of pregnancy on sleep characteristtcs have been studied m a survey reported by Schweiger [l] A maJortty of the 100 women contacted m that survey reported havmg sleep disturbance with generally more awakenings and less total sleep tune than normal. Sleep disturbance was reported as most pronounced during the third trimester of pregnancy. This generally agrees with the findings of Karacan et al [2,3] using EEG momtormg during pregnancy Prehmmary results from the present study have previously been described by Karacan et al [4]. *This research was funded by NIMH Grant No 15900-04 $ Departments of Soctology and Psychtatry, Umverstty of Florida Gamesvtlle. Florida 6 Department of Psychiatry. Umverstty of Loutsvtlle, Loutsvtlle. Kentucky

Recently Kales et al [S] presented a summary of results obtained from quesuons concerning sleep disturbance from a representattve sample of 1000 households m the 1973 Los Angeles metropolitan survey They found that 32.2% of the respondents had complamts of msomma mcludmg 22 8% complammg of awakenings durmg the night, 144% havmg trouble falling asleep, and 13 7% experiencmg early morning awakenings Some mdivtduils obviously reported havmg more than one type of sleeping difficulty Females had greater mcidence of sleep disturbance than males, and problems increased with age In general these two surveys are mutually supportive, but dtfferences exist primarily with respect to the breakdown of rates among the various categories of sleep disturbance One of the most comprehensive surveys of normal sleep patterns was done by McGhie and Russell [6] Questionnaires were obtamttd from 2446 Scottish subJects, rangmg in age from 15 to 75+ years, from the Dundee and Glasgow areas, relatmg to tune different questions on sleep and sleep-related charactertstics. The study concentrated on the effects of age on normal sleep patterns but was also concerned with dtfferences due to sex and so&l class. Most variables showed large dflerences due pnmarlly to age and secondarily to sex, with older females exhibiting the greatest disturbance of sleep and similarly the highest rate of hypnotic drug utllu;8tion Although caution must be exercised m the mterpretatlon of SubJectlvely reported effects on sleep [7], these measures have been shown to be useful, both as independent descnptors of sleep and as mformation complementary to that provided by the obJectively measured EEG parameters [8-101 Data from our laboratory has shown that among normal subjects a high correlation exists between certain tradttionally important sleep EGG parameters and their SubJective counterparts, especially sleep latency, total sleep time, and number of awakenings [ll] Thus, despite the questionable vahdtty often attributed to certain non-EEG estimates of sleep [12], it does not necessarily follow that all non-EEG evaluations are mvahd On the contrary, the evtdence further suggests that subJective reports provide unique mformation

239

I

240

KARACAV er trl

such as suficlency and restfulness of sleep, unattamable m any other manner Disturbance of sleep 1s probably the most common sleep disorder, and may be one of the most common of all medical disorders Patients often complam of this problem m hlstorles obtained at hospitals and outpatient chmcs Despite the fact that msomma apparently affects large segments of the population to varying degrees, there exists very httle evidence of Its extent Furthermore, the limited available evldence should undoubtedly be replicated m different population areas Thus, there IS an evident need for further eptdemlologlcal studies of sleep and its chsorders It IS m this sense and with this spirit that questions pertaming to sleep were integral components m the questionnaire developed for the Florida Health Study Program, the results of which are the SubJect of this paper METHOD

Destgn of the survey

Data for this study were obtamed from a health survey conducted durmg 1969-1970 m Alachua County, Florida, with an area of 982 square miles and a 1970 population of slightly greater than lOOOC0 The major share of this population consists of approximately 75 000 people hvmg m or near the central city of Gamesvllle, a rapidly growmg education, medlcal, and technologcal center m north-central Florida The survey consisted of a comprehensive 317-item interview schedule designed to elicit mformatlon about five general categories of variables mcludmg (1) demographic data and social history, (2) famlhal and mterpersonal relations, (3) religion, race, and other soc~ological variables, (4) physical and mental health, and (5) attitudes toward utdlzatlon of health care services Two previous reports of results from this survey are the pubhcatlons of Warhelt et al. [13,14] Imtlally, a random sample of 2333 adult respondents from the county’s 37000 households was selected to be mtervlewed, using tables by Klsh [15] to assure randomness After a pretest involving 322 of these respondents, another 1645 mdlvlduals were interviewed followmg revision of the interview schedule, making a total of 1967 interviews, or almost 85% of the intended sample (8 8% refused and 69% could not be contacted despite repeated attempts) Results reported here are for the 1645 mdlvlduals m the mam survey An analysis of soclodemographlc characterlstlcs of the sample m comparison to the 1970 census reports showed it to be representative of the county’s adult population Vurruhle~

In this paper we will present questionnaire data describing evidence of sleep problems across vdrlous segments of the population Our focus 1s on five soclo-demographic variables (I ) race, (2) sex, (3) age, (4) socloeconomlc status (SES) and (5) marital status All except SES were lnchvldual Items on the questlonnalre SES was obtained as a composite of an mdlvldual’s education, occupation, and income, usmg a method suggested by the US Bureau of the Census [ 163 As calculated, the SES scores reflect an mdlvl-

dual’s relative standing wlthm our sampled population (le ilot relative to the national census data) SIX mdlvldual questionnaire items provided mformatlon concerning problems associated with sleep 1 How often do you have trouble wnh sleepmg” Possible ratings (a) never (b) seldom. (c) sometimes (d) often. (e) all of the time 2 Do you have any trouble getting to sleep and staqmg asleep? Possible ratmgs (a) never. (b) sometlmes. (c) often 3 Are you ever bothered by nightmares (dreams which frighten you? Possible ratmgs (a) never (b) sometlmes. (c) often 4 How often have you used pills to help you get to sleep? Possible ratings (a) never, (b) seldom. (c) sometimes (d) often (e) all the time 5 What kmd of trouble do you have” Possible ratings (a) fallmg asleep, (b) staymg asleep, (c) not enough sleep (d) too much sleep. (e) wakmg up too early, (0 more than one category 6 Did you have trouble this week” Possible ratmgs (a) yes, (b) no

Questions 1. 2. 3. and 4 were asked of all respondents Questions 5 and 6 were only asked if the response to question 1 was other than “never ” Respondents rated almost all of these items as requested There were very few instances of “don’t know,” “not answered,” or “not applicable ” Percentage breakdowns of various demographic characteristics for subjects in our survey were compared to snnllar breakdowns from the 1970 census Census categories were not defined the same as ours, and converSions had to be made for comparablhty It was evident that out sampling was especially representative of the county’s race and sex dlstrlbutions On the other hand, single and younger mdlviduals (le the college population) were somewhat undersampled No comparisons were made for socmeconomic status RESULTS

The data were summarized at two levels At one level we studied the marginal effects of each SOCIOdemographic variable on the various sleep variables The second level of analysis evaluated the Joint slmultaneous action of all soao-demographic variables on each sleep variable These more penetrating analyses investigated the posslbtlltles that marginal effects may not represent the true contrtbutlon of sociodemographic variables, certainly not if they mteractlvely relate to sleep charactenstlcs The marginal analyses. illustrated for one of the items on Fig 1, were all performed using computer program “Crosstabs” m the SPSS program system [I73 Slgmficance was based on the x2 analysis of frequencies m the two-way tabulations Although the sampling m this study was representative of the population m Alachua County so that the marginal analyses provide estimates of selfreported sleep characterlstlcs for each separate SOCIOdemographic component. the analyses fall to show how the results might change when consldermg combinations of variables For example m Fig 1 we show that “trouble with sleeping” was directly related to being black, being female, in an older age group, at a low socloeconomx status. or being widowed, separated, or divorced What we cannot determine from

Prevalence of sleep dlsturbance

these data IS the effect of bemg all of these thmgs Might older divorced black females m a low SES category experience more, or possibly less, trouble than would be predlcted from each separate category? A further m-depth analysis of the data was made using a program for the analysis of categorical data by hnear models [I81 A feature of the program IS that data must be expressed in a two-dlmenslonal array subpopulations (rows) by categones of response (columns) Each cell of the matnx would rep resent the number of lndlvlduals m a particular subpopulation giving a specific rating. After dlchotomlzlng age, SES, and marital status mto low and high categories. the subpopulatlons consisted of all 32 combmatlons of the followmg Race

White vs black Male vs female 139 vs 240 Age SES 139 vs 240 Marital Single vs other Response categories conslsted of the ratings g1ve.n to the questions concerning sleep, and quantltatlve scores were assigned to these ratings never = 0, seldom = 1, sometimes = 2, often = 3, and all the time = 4 The boundaries between low and high categories for age and SES were chosen to provide an empmtally optimum &vlslon based on the results of the first-stage analysis Marital status was combined as defined above only after findmg retrospectively that “single” could be compared to all “other” as a conglomerate, since married mdlvlduals did not differ from widowed, separated or divorced mdlvlduals m this analysis Sex

How often do you have trouble with sleepmng?

Figure 1 shows that more than one-third of the respondents clanned to have trouble with sleeping Tot01

1

646

RZZRwq x2-682 PC005

White Block Mole Female

70.0 I

603

be 18-19 20-29 3D-39

62 x2=63 31

40-49

p
50-59 60-69 a70

Sex 26

40-49

50-59

;

M F

60-69

‘:

a70

‘:

~orseklom~scnwtmEs

t_3ly~Qy”

Fig 2 How Often Do You Have Trouble With Sleepmg7 InteractIon of percentage responses due to age and sex

(22 O”/, “sometimes,” and 53 4% “often” or “all the time”) Note that we have compressed the response categories here and on Figs. 2 and 4 to attain greater umfornuty and for convenience The rate was shghtly higher for black respondents, higher for females, dlrectly related to age (6.2% of mdlvlduals 19 years of age or under had trouble “often” or “all of the time” as compared to 25 9% of lndlvrduals 70 years of age or over), inversely related to socloecononuc status (22.0% of mdlvlduals m SES level 1 had trouble “often” or “all the time” as ,compared to 9 0% of mdlvlduals at SES level S), and was considerably lower for single or married as cowpared to widowed, separated, or divorced mdlvldurdls The linear models anal@ largely substantiated these findings, but suggest&d m addrtlon an mteracbon effect between sex and age Figure 2 shows that the rates are approximately equal for both sexes through age 29, but for all older age ranges females have greater dlficultles Do you have trouble gettzng to sleep and stayuly asleep? This item gave almost ldentlcal percentage breakdowns as described above fgr sleepmg trouble In general, but with slightly hlgber rates of difficulty (not tabulated) Altogether, nearly half of the respondents clauned to have trouble geitmg to sleep and staymg asleep (30.8% “sometimes” #d 13 8% “often”) Except that there were no dlfferenqes due to race, the socmdemographlc vanables had the same patterns of slgmficant differences as observed for the previous item Are you ever bothered by nvhtmares?

O-19

20-39 40-59

x2=5023 p
60- 79 80-100 Smgle Morrled WID/SEP/DIV ONever or seldom

Age

241

x2.4701 p
DOften or oil the time

Fig 1 How Often Do You Have Trouble With Sleepmg? Percentages of responses m total and for each of five margmal soclo-demographlc categories

Figure 3 shows that sh&tly less than one-fourth of the respondents were ever bothered by nightmares (20 2% “sometmes,” and 2.4% “often”) There were no slgmficant hfferences due to race, age, socloeconomic status or manta1 status,; but a shghtly higher rate for females (24.0% “sometimes,” and 3 2% “often”) than for males (15 5% and 14x, respectively) Also there was an mdlcatlon of a shghtly decreasing mcldence with mcreasnig SES $tatus The linear models analysis Indicated a highly slgmficant interaction between sex and age Clearly, females have relatively

242

I KARACAN et nl

1

Total

773

R202a

24

Age

sex

18-19

M F

48 26

20-2s

“:

09 34

30-3s

;

00 42

40-4s

;

IO 43

50-5s

“;1

IO 24

60-69

;

a70

y

40 35 INever

WSorneh

Fig 3 Are You Ever Bothered by NIghtmares? Percentages of responses in total and with respect to the mteractlon between age and sex

greater problems through age 48, but then males are bothered by mghtmares at least to the same degree. How often have you used prlis to help you get to sleep? Figure 4 Indicates that approxtmately 10% of the respondents use sleepmg ptlls (7.1% “somettmes,” and 3 4% “often” or “all the tune”). The rate of utthzatlon

did not differ slgmficantly by race or socmecononuc status, which were somewhat surpnsmg results. Females had a higher rate than males (13 6% of females using sleepmg pills sometimes or more frequently as compared to 6.7% of males) The rate of utlhzatlon Increased steadily and dramattcally wtth Increasing age (13% of mdlvlduals 19 years of age

Total Age 18-B

Sex M F

20-2s

;

30-3s

M F

16 28

M

F

IO 58

l” F

22 56

;

69 75

50-5s 60-69

What kmd of sleepmg trouble do you have9

The most common complaint by far was falling asleep, as shown m Fig 5 containing responses for only those mdlvlduals expressing the presence of some kmd of sleeping problem More than half of the respondents complamed of trouble falling asleep, and 19 9% complained of trouble m more than one category The second highest mdlvldual category was staying asleep (15 3%) followed by msufficlent total sleep time (6 5%) and early awakening (3 1%) An additional five respondents reported havmg too much sleep There were no slgruficant differences m rates due to race, sex, socloeconomlc status, or manta1 status, but mterestmg differences m the various age groups Respondents 19 years of age or under had a higher percentage m the “fallmg asleep” category than any other age group, respondents between 20-39 years of age had proportionately higher percentages m the “not enough sleep” category, and respondents 40 years of age and older were proportionately higher m the “staying asleep” and “wakmg too early” categories The complamt of msufficient total sleep time for mdlvlduals m their 20s and 30s may be largely due to constraints Imposed by then life style The increasing dtfficulty of staymg asleep and waking too early for older respondents IS very likely attributable to the aging process, but may also be a reflectlon of their takmg problems to bed with them That the category “not enough sleep” may be related to hfe style 1s also indicated by the (nonstgmficant) tendency for higher percentages m this category for mdlvtduals m the upper SES levels and for single mdlvlduals Did you have sleepmg trouble this week?

13 2.3

40-4s

or younger use sleepmg ~111ssometlmes or more frequently as compared to 18 37; of mdrvlduals 70 years of age or older) With regard to manta1 status, single mdlvlduals had the lowest rate (5 872 “sometimes.” “often,” or “all the time”), whereas wldowed, divorced, or separated mdlvlduals had nearly three times the rate (15 6%). Agam, there was a highly significant mteraction between age and sex Figure 4 shows that the differences were mmrmal through age 29 (recall Fig. 2), but then females had much higher rates than males, especially m the 40-49 age bracket

Indlvlduals responding to this Item had comparable rates across all soao-demographic categories. Altogether, 52% of the 931 mdlvlduals responding to this question had sleeping trouble wlthm the current week

AW l9-I9 20-29 30-39 40-49

cdl-69

60-69 270

Fig 4 How Often Have You Used Pdls to Help You Get to Sleep’, Percentages of responses m total and with respect to the mteractlon between age and sex

Fig 5 What Kind of Trouble Do You Have7 Percentages of responses

m total

and with respect

to age

243

Prevalence of sleep disturbance DlSCUSSlON

Our survey based on a random sample of 1645 adult mdrvtduals from Alachua County, Florida, revealed patterns of sleep problems dtffermg among socio-demographic categories More than one-thud of all respondents reported havmg trouble with sleeping The problem IS greater for older non-single females at a low SES level, but IS unrelated to race Slm~larly, the specific problems of getting to sleep and staying asleep gave nearly parallel results, partially reflectmg the fact that almost 900/, of the respondents reporting sleepmg trouble of any kind had trouble getting to sleepand staymg asleep. Problems with nightmares were greatest for young females at a low SES level The use of pills to aid m gettmg to sleep 1s highest among older white females, and widowed, separated or divorced mdtvrduals The various types of sleeping trouble were equivalent across most soclo-demographic variables, drffermg only due to age Assuming that mdivtduals clamnng more than one category were distributed proporttonately across the mdivrdual categories, we would find the followmg breakdown falling asleep (68 5x), staying asleep (19 Ox), not enough sleep time (8 lx), waking early (3 9x), and too much sleep (0 5%) The relatronship between age and type of sleepmg trouble could be mdtcattve of life-style drfferences For example, mdivtduals between 20 and 39 years of age had proporttonately high rates of msufficrent total sleep, possibly reflecting then active life style. Waking too early was most pronounced among mdtviduals 40 to 49 years of age, and lowest among all younger mdividuals Respondents 50 years of age and older tended to be drsproportronately more bothered by difficulty staymg asleep The problem of sleep latency. was highest among the youngest age group Further insight mto problems associated wtth sleep could be obtained from crosstabulatlon between the various sleep variables For example, companng kinds of sleeping trouble with utthzation of sleeping pills, we found that frequent use (often or all the time) was highest for mdrvtduals reporting more than one category (9 8x), intermediate for mdtvtduals having trouble falhng asleep (5 2%) staying asleep (3 5%) and wakmg)oo early (3 4%) and mmimal for mdtviduals complaunng of not enough sleep (1 7%) or too much sleep (0 0%) Differences were found between the results of our survey and those of the Los Angeles survey reported by Kales et al [5] We found that 58 0% of the mdtvtduals reported having trouble with sleeping at least seldomly, whereas the latter study found a total of 322% having complamts of msomma. Note that d we had excluded the “seldom” responses, our rate would have been 35 4”/ or very similar to the Los Angeles results Compansons between the two surveys with respect to types of sleeping dtfficulty are harder to construct The most obvtous difference was the relatrve predominance of sleep latency m our study as opposed to sleep maintenance m the Los Angeles survey Results of the questronnane survey reported by McGhre and Russell [6] generally agreed wtth ours, mdicatmg sex- and age-related sleep differences, especially with respect to the use of hypnotics However, the samples of respondents m the two communl-

ties were not representative of the population, and furthermore several of the questtonnane items could not be interpreted analogously to those m our survey Then utrhzation of sleep length, sleep latency and time of awakening as measures of sleep sufficrency, difficulty falling asleep and early morning awakenmg respectively may be questioned because of known dtfferences among normal mdivtduals m these respects c71 Discrepancies among survey results may be due to a number of causes Both the questions asked and then appearance m context figure prominently in assessing differences m survey results, but few if any attempts have been made to establish their rehabthty and validity Regional and seasonal dtfferences may be responsible for a certain degree of variation among surveys Finally, sampling &fferences may lead to dtfferent socio-demographic representattons. For these reasons, replication of SUrV&yS in different commumties and at different points tn time may yteld contrasting patterns of results We feel that sample surveys can be utihzed to provtde useful results concernmg sleep problems m a population, and propose that standardized methodology should be developed SO that comparable mformatron would be obtamed Rehabthty and validity of the test mstruments can be assessed by parallel laboratory monitoring usmg objective measures Based on experience m our laboratory usmg patients (insomniacs) and normal sleep subjects, the data suggest that although normal mdtviduals are able to accurately descnbe mghts of both good and poor sleep, msommacs have more difficulty and tend to present a biased assessment [20,21 J If such evidence is supported and m&tamed, then efficacy of the survey approach may be SubJect to question Thus, smce the survey approach does have tts hmrtatron and even though it IS an excellent tool for obtammg statrstics on prevalence and type of sleep disorders, the ultimate quantrfication of sleep patterns must presently rely on laboratory momtonng of EEG parameters However, m spite of its hmitations, the survey approach provides extensive mformatton otherwise unobtamable by laboratory analysis, thereby potenttatmg differentiatton and class&cation of sleep disorders Further, these data allow us to formulate Ideas and provrde further hypotheses that can be studied more rigorously m the laboratory By the very nature of its ability to study p6Opl6 m then natural habttat, the sleep survey allows us to study the mdtvtdual rn situ and does not suffer from the disadvantage of “artlficlallty” often attributed to any laboratory assessment It is thus felt that the use of surveys merits further attention The ultimate critenon as to what constitutes a good night’s sleep mu): be a sublectrve one Erratum-Figure 1, age 40-49, “somettmes” response should read 23 8%

REFERENCES

1 Schwelger M S Sleep disturbance In pregnancy A subjective survey Am J Obstet Gynecol 114, 879, 1972

244

I KARACANer al

2 Karacan I. Heme W. Agnew H W. Wdhams R L, Webb W B and Ross J J Characterlstlcs of sleep patterns durmg late pregnancy and the post-partum periods Am J Obsret Gyttecol 101, 579, 1968 3 Karacan I, Wdham‘s R L, Hursch C J , McCaulley M and Heme M W Some lmphcatlons of the sleep patterns of pregnancy for postpartum emotional dlsturbances Br J Psychtat 115, 929, 1969 4 Karacan I, Warhelt G G, Thornby J I. Schwab J J and Wtlhams R L Prevalence of sleep disturbance m the general population (Abstract) In Sleep Research (Edited by Chase M H. Stern W C and Walter P L ) 2. 158, 1973 Los Angeles Brain Information Servlce/Bram Research Institute, Uruverslty of Cahforma 5 Kales A, Blxler E 0, Leo L A, Healy S and Slye E Incidence of msomma m the Los Angeles metropohtan area Paper presented at the Annual Meeting of the Assoclatlon for the Psychophyslologcal Study of Sleep at Jackson Hole, Wyommg, 1974 6 McGhle A and Russell S M The subJective assessment of normal sleep patterns J Ment Scz 108. 642. 1962 7 Dement W C . Zarcone V P, Hoddes E , Smythe H and Carskadon M Sleep laboratory and chmcal studies wth flurazepam In The Bentodmzepmes (Edited by Garattml S, Mussml E, and Randall L 0 ) pp 599-611 Raven Press, New York. 1973 8 Monroe L J Psychologcal and physlologcal differences between good and poor sleepers J Abnorm Psychol 72, 255. 1967

9 Hartmann E, Baekeland F and Zwdlmg G R Psychological differences between long and short sleepers Arch Gen Psychlat 26 (May), 463, 1972 10 Stonehlll E and Crisp A H Aspects of the relatlonshlp between sleep, weight and mood Psychother Psychosom 22(2). 148, 1973

I1 Thornby J I, Karacan I, Beutler L E , Booth G H , Anch A M, Wdhams R L and Sahs P J Once more

12

13

14

15 16 17 18

thrice Objective-subjective correlations m sleep Paper presented at the Annual Meeting of the ASSOCIatlon for the Psychophyslologlcal Stud! of Sleep Jackson Hole. Wyommg. 1974 Weiss B L, McPartland R J dnd Kupfer D J Once more The inaccuracy of non-EEG estlmdtlons of sleep Am J Psvchrar 130 (Nov) 1282. 1973 Warhelt G J, Holzer C E III and Schwab J J An analysis of social class and racial differences m depresslve symptomatology A community study J Hlth Sot Behac 14 (Dee), 291 1973 Warhelt G J. Schwab J J Holzer C E III dnd Nadeau S E New data from the South on race sex age and mental Illness Presented at the Annual Meeting of the American Sociological Association in New York, NY, 1973 Klsh L Surte) Samplmg Wiley. New York, 1965 United States Bureau of the Census US Census of the Population SubJect Reports Socioeconomic Status PC(Z)-SC. 1960 Nle N H. Bent D H and Hull C H SPSS Statrstrcul Package for the Social Scrences McGraw-Hill New York, 1970 Forthofer R N, Starmer C F and Grizzle J E 4 Program for the Analysu of Categor lcal Dara bt Linear Models Chapel Hill Umverslty of North Carohnd In-

stitute of Statistics Mlmeo Series No 604. 1969 19 Wllhams R L , Karacan I, and Hursch C J Elecrroerlcephalography (EEG) of Human Sleep cattons Wiley, New York. 1974

Clrmcal Apph-

20 Karacan I and Wdhams R L Insomnia Old wme m a new bottle Psvchtat Q 45. 274 1971 21 Blxler E 0, Kales A. Leo L A dnd Slye T A comparison of SubJective estimates and objective sleep laboratory findings m msommac patients Paper presented at the Annual Meeting of the Association for the Psychophyslologlcal Study of Sleep at San Diego. 1973