435

Original Research Communications-general

Caloric intake necessary for weight maintenancein anorexia nervosa: nonbulimics require greatercaloric intake than bulimics13Walter H Kaye, MD, Harry E Gwirtsman, MD, Eva Obarzanek, PhD, Ted George, MD,

David C Jimerson, MD, and Michael H Ebert, MD

ABSTRACT In the past decade, patients with anorexia nervosa have been subdivided by thepresence or absence of bingeing-and-purging behavior. Psychologic, physiologic, and premorbid weightdifferences have also been discovered between these subgroups. We now report that nonbulimic

anorectics required 30-50% more caloric intake than bulimic anorectics to maintain a stable weight.

This difference in caloric intake was independent of phase of illness; it was present at low weight andat intervals after weight restoration. Subjects were closely supervised on an inpatient hospital ward

so that they could not binge or purge. Motor activity did not appear to explain these alterations incaloric requirements. Such differences in caloric intake could be trait related or a consequence ofmany years of starving or bingeing behavior. These findings are clinically relevant for advising eating

disorder patients of caloric requirements necessary to maintain a normal weight. Am J C/in Nuir1986;44:435-443

KEY WORDS Anorexia nervosa, bulimia, caloric intake, energy metabolism

Introduction

Patients with anorexia nervosa can be sub-divided into two groups by appetitive behav-ior those that fast and those that binge (1-6).These subgroups of anorectics are also psy-chologically different. Bulimic anorectics havebeen found to be more outgoing, more im-pulsive, more mood labile, and more sexuallyactive than nonbulimic or food-restricting an-orectics. We have recently reported (7) thatthese two groups, after weight recovery, differin turnover of CNS serotomn.

Several investigators (1, 3) have reportedthat bulimics have a greater premorbid bodyweight than do nonbulimic anorectics. Oneexplanation is that there are differences in ef-ficiency of energy metabolism between thesetwo groups of patients such that bulimic an-orectics gain weight more easily than nonbu-limic anorectics. It has been suggested that an-

imals can alter the efficiency with which theyutilize energy contained in food (8), althoughsuch studies have been questioned (9).

Several investigators have provided evi-dence that some human adults, most notablyobese or formerly obese individuals are moreenergetically efficient than lean controls in

From the Department of Psychiatry (WHK), Uni-

versity of Pittsburgh School of Medicine, Western Psy-chiatric Institute and Clinic, Pittsburgh, PA; Laboratoryof Clinical Science (HEG, EO, TO, DCJ), National Insti-tute of Mental Health, Bethesda, MD; and Departmentof Psychiatry (MHE), Vanderbilt University, Nashville,TN.

2This study was performed at the Intramural programof the National Institute of Mental Health, Bethesda, MD.

3Address reprint requests to: Walter Kayc, MD, West-ern Psychiatric Institute, University of Pittsburgh, 3811O’Hara Street, Pittsburgh, PA 15213.

Received September 16, 1985.Accepted for publication March 13, 1986.

The American Journal of Clinical Nutrition 44: OCTOBER 1986, pp 435-443. Printed in USA© 1986 American Society for Clinical Nutrition

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

436 KAYE ET AL

* p <0.05, bulimics vs nonbulimics.

terms of their metabolic response to thermo-genie stimuli such as glucose (10), mixed meals(11), postprandial exercise (12), and a ther-mogenic drug (13). Thus, although contro-versial, the possibility remains that differencesin efficiency of energy utilization exist amonghuman adults.

This study was done to test whether caloricintake would differentiate between a popula-tion of anorectics that either fast or binge andwhether this difference was independent ofphase of the illness. Such a finding might havewider implications for understanding weightregulation in humans as well as the relationbetween appetite and energy-metabolism ef-ficiency.

Methods

Subjects

All subjects were hospitalized in clinical research unitsof the National Institutes of Health Clinical Center, Na-tional Institute of Mental Health. Subjects gave informedconsent for the study. Patients who met DSM-III criteriafor anorexia nervosa(14) were studied during three phasesof illness: 1)11 anorectics (Table 1) were studied after 6mo or longer of being continuously at <75% of averagebody weight (underweight anorexia nervosa); 2) the same11 underweight anorectics (Table 2) were restudied 3-4wk after correction of weight loss (recently weight-restored);

and 3) a separate group of 9 women who had been un-derweight with anorexia nervosa (Table 3), but who at thetime of the present study had been weight recovered forat least a year (long-term weight-restored). A tenth subjectin the third group (number 11) was also studied at lowweight and after short-term weight restoration. Long-termweight-restored women had been able to maintain weightrecovery for a mean (±SD) of 32 ± 12 mo (with a rangeof 1 to 10 yr). Control subjects (Table 4) consisted of 11healthy women who had no medical, neurologic, or psy-chiatric problems.

The anorectic patients were separated by style of ap-petitive behavior into groups based on history obtained

from the patient and, when possible, from relatives. Non-bulimic anorectics had lost weight by restricting caloricintake; these subjects never binged. Bulimic-anorectic pa-tients engaged in bingeing behavior and all vomited and/or used laxatives after bingeing, In this study, patients whoused laxatives(n = 5)but binged only a few times a month(subjects number 4, 5, 12, 15, 17) were grouped with bu-limic anorectics. It is unsettled as to whether those whoengage in vomiting or laxative abuse without bingeingshould be considered in the bulimic contingent (1) orshould be considered fasters (3, 4).

Caloric data

Subjects were restrictedto a locked ward for control ofenvironmental temperature, motor activity, and food in-take. Following admission, underweight anorectics werenutritionally stabilized at a stable low weight that did notvary more than ±1.0 kg for 4-6 wk until they began theweight-gaining phase of the program. Refeeding and weightgain were accomplished by a modified behavior-modifi-cation program: no medication was used. The target weight

TABLE 1Anorexia nervosa patients at low weight

SubjectNumber of daysat �ab1e weight

Mean weight (kg)Mean ± SD

Daily ealoric intakeMean ± SD

Caloric intake perbody suiface area

kcdmi�

Bulimics

la

2a3a4a5a

MeanSD

12

971211102

39.9 ± 0.2236.6±0.4032.6±0.2230.8 ± 0.2832.5±0.09

34#{149}5*3.7

925 ± 1671168±268713±207963 ± 228601± 99

874222

642859575789459665*161

Nonbulimics

6a7a8a9a

l0ailaMeanSD

9698

2119126

31.7±0.1025.5±0.4833.4±0.2429.9±0.0728.6±0.1327.9±0.21

29.5*2.8

1150±1861163±311939±198927±1711029± 1891015±248

1037101

8911038751792887868871*99

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

*p<OOl

ENERGY METABOLISM IN ANOREXIA NERVOSA 437

TABLE 2Anorexia nervosa after short-term weight restoration

Number o(days Mean weight (kg) Daily caloric intake Caloric intake perSubject at �ab4e weight Mean ± SD Mean ± SD body surface area

kcai/m’

Bulimics

lb 15 49.5±0.17 2130±330 13572b 28 49.5±0.42 2001±651 12913b 11 47.4 ± 0.24 2068 ± 538 14074b 29 44.1±0.45 1451±336 1022Sb 9 47.6±0.28 1807±259 1181

Mean 18 47.6* l89lt l251tSD 9 2.2 274 154

Nonbulimics

6b 13 46.1±0.17 2399±268 1589lb 22 43.5 ± 0.23 2645 ± 428 18768b 39 40.6±0.20 2197±324 16279b 41 41.6±0.14 2592±637 1920lOb 28 41.7±0.24 2203±364 1620Ilb 20 46.5±0.21 2214±223 1527Mean 27 433* 231St 1693tSD 11 2.5 204 163

* p <0.05, t p < 0.01, bulimics vs nonbulimics.

thatwas defined as reaching short-term weight restoration program during short-term weight restorationso that wewas the weight where women with secondary amenorrhea could get them to eat enough food to maintain a stablemight be expected to menstruate (15, 16). After weight weight. The long-term weight-restored and the controlrestorationwas completed, anorecticswere again asked to subjects were brieflyhospitalized under similar conditions.maintain a stabletargetbody weight (±1.0 kg) for4-6 wk. Subjectswere allowed only three45-mm meals/day andAnorectic patients continued in the behavior-modification three IS-reinsnacks.No patientson the ward were allowed

TABLE 3Anorexia nervosa after long-term weight restoration

Number of days Mean weight (kg) Daily caloric intake Caloric intake perSubject at stableweight Mean ± SD Mean ± SD body surface area

kcoJ�

Bulimics

12 5 42.5±0.08 715±231 53413 3 39.6±0.38 1055± 74 78714 8 60.5 ± 0.49 1040 ± 343 60515 2 49.9±0.21 1125± 64 71716 4 49.9±0.32 1256±191 81517 4 59.1±0.26 1373± 191 822Mean 4 50.3 1094* 713SD 2 8.5 225 120

Nonbulimics

18 2 55.2±0.18 1935±599 125619 4 53.9±0.14 1423±333 90111 3 45.6±0.20 1595±198 110820 3 42.0 ± 0.23 1407 ± 207 1027Mean 3 49.2 1590* 1073*SD 1 6.4 245 149

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

438 K.AYE F� AL

TABLE 4Healthy control women

SubjectNumber of daysat stable weight

Mean weight (kg)Mean ± SD

Daily caloric intakeMean ± SD

Caloric intake perbody surface area

kcd/m�

21 4 60.2 ± 0.08 1428 ± 485 84022 5 70.2 ±0.27 1572 ±650 85023 2 46.1±0.21 1795±233 121324 4 59.3 ± 0.04 2421 ± 401 143325 5 49.6±0.14 1687± 150 112526 7 61.0±0.37 1866±308 111127 8 59.8±0.42 1389± 172 82228 6 50.0±0.30 1860±367 122429 5 53.2±0.19 1767±204 112530 6 54.1±0.60 1334±293 88931 3 65.4±0.15 1858±285 1074Mean 5 57.2 1725 1064SD 2 7.3 304 194

to have food in their rooms. All subjects were observed24 h/day on the ward, including mealtime and bathroomvisitsso they could not secretly binge or vomit. All foodwas ordered from the hospital kitchen and caloric contentdocumented before being given to the patients. After theuneaten food was returned to the kitchen and reweighed,an estimate of daily caloric intake was made. In an in-dependent study, the accuracy of the method of caloricestimation used in this report was checked (17). Clinicalcaloric estimates were 102.2 ± 2.2% of the values foundby chemical laboratory calorimetry, a nonsignificant dii’-ference.

Low-weight anorectics were often dehydrated on ad-mission. Thus weight and calorie data in the interval afteradmission (usually the first 7-14 days of hospitalization)were discarded. During this initial interval of time, thedietitian worked with the subjects to adjust caloric intakein order to establish the amount necessary to maintain astableweight For this study, we used the longest sequentialnumber of days where weight remained within ±1.0 kgbut where there was no overall weight gain or loss. Todetermine that there was no overall weight gain or lossduring this period of time, a linear-regression coefficientwas calculated (daily weight versus number of days) (18,19). The period of days was adjusted until the regressioncoefficient was nonsignificant (p> 0.05), indicating thatthere was no positive or negative linear trend in weightover this time interval. After the patients attained theirtarget weight, daily caloric consumption for the short-termweight-restoration phase was similarly calculated. Short-term weight-restored anorectics required an interval ofabout 4-14 days after target weight was achieved to adjusttheir caloric intake so that they were able to maintain astable weight. Data from this adjustment phase was dis-carded.

Long-term weight-restored anorectics and normal con-trolswere only willing to remain locked on the ward, undersimilar conditions, for relatively brief periods of time (fewerthan 8 days). Thus much less data are available for thesegroups. Although 12 long-term weight-recovered subjectswere studied, 2 lost weight during theirinpatient stay andso are not included. All normal controls were able to re-

main within 1 kg of their admission weight during this

study.

Motor activity

Motor activity was automatically and continuously re-corded for 24 h/day for 3-5 days by methods previouslyreported (20, 21). Motor activity was measured by anacceleration-sensitivedevice with a solid-statememory thatstores data on the number of motor movements (22). Themonitors were attached to a belt around the waist of thesubjects. Two monitors were used throughout the study.These monitors, initially calibrated to be equal to eachother, maintained a variance of< 7% throughout the study.Motor activity was not obtained on one long-term weight-recovered anorectic and one control.

Caloric correction for weight

Because the subject groups differed in weight, somemethod of comparing caloric intake corrected for differ-ences in weight was necessary. Several methods of correc-tion are widely used (23-27) but there is no clear agreementon the best method. We expressed total daily caloric intakeinterms of body weight,body mass index, (weight/height2),and body-surface area (28).

Data analysis

Bulimic and nonbulimic anorectics were comparedwithin each phase of study by an independent two-tailedI test. Daily caloric intake per square meter of surface areawas covaried with daily counts of motor activity by one-way ANOVA (29).

Results

Bulimic anorectics (28 ± 3 yr) were signif-icantly older (p <0.05) than the nonbulimicanorectics (23 ± 3 yr) in the group studiedwhen underweight and after short-term weight

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

SS

tSS

2000

1500

1000

500�

C,’

E

A

A

ft

S

+0

SOS

-�-*

0

0

-0- * *

0

0

LOW WEIGHT SHORT TERMWEIGHT

RESTORED

HEALTHYCONTROLWOMEN

ENERGY METABOLISM IN ANOREXIA NERVOSA 439

recovery, although the difference in mean agesof these two groups was only 5 yr. Long-termbulimic and nonbulimic anorectics were ofsimilar ages (26 ± 5 vs 26 ± 6 yr). The agesof the healthy control women (23±4 yr) weresimilar to the anorectics. All groups of ano-rectics had a similar duration of illness priorto study. In the group studied at low weightand after short-term weight restoration, bu-limic anorectics had been ill for 96 ± 48 moand nonbulimic anorectics for 84 ± 28 mo.For the long-term weight-restored anorectics,bulimics had been diagnosed as having an-orexia nervosa 112 ± 51 mo prior to studyand nonbulimic anorectics for 95 ± 49 mo

prior to study. Heights were similar betweenall subgroups of anorectics and controls.

Low-weight patients with anorexia nervosamaintained a stable weight for 6-21 days(Table 1). Bulimic anorectics were significantlyheavier than nonbulimic anorectics but con-sumed a similar number of calories/day. Whencaloric intake was corrected for body surfacearea, bulimic anorectics consumed signifi-cantly fewer calories/day than nonbulimic an-orectics (Figure 1). Bulimic anorectics weresignificantly more active than nonbulimic an-orectics (Figure 2).

After short-term weight recovery (Table 2),bulimic anorectics were still significantly

0

0

0

-**

0

#{149}NON BULIMIC ANORECTICS

o BULIMIC ANORECT1CS

A HEALTHY CONTROL WOMEN

#{149}

0 #{149}

LONG TERMWEIGHT

RESTORED

PATIENTS WITH ANOREXIA NERVOSA

FIG 1.Daily caloric intake corrected for body surface area at phases of anorexia nervosa. Anorectics are separatedby appetite-relatedbehavior into nonbulimic (open circles) and bulimic (closed circles) subgroups in each phase.Healthy control women are indicated by triangles. All subjects maintained a stable weight (± 1.0 kg) during caloricmeasurement Subgroups are compared by two-tailed group t test, *p <0.05, � <0.01.

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

C

8

>I-U

-J

U

Cl)>-

a-

6000-

5000-

4000-

3000-

2000-

1000-

0

-*

0.

.

.

S

#{149}NON-BULIMIC ANORECTICS

o BULIMIC ANORECTICS

A HEALTHY CONTROL WOMEN

0

-0- *

0

$-o

#{149}

S

0

$

-a-#{149} 0

#{149} 0

0

0

A

A

-a-A

A

LOW WEIGHT SHORT TERMWEIGHT

LONG TERMWEIGHT

iRESTORED RESTORED

HEALTHYCONTROLWOMEN

440 KAYE ET AL

PATIENTS WITH ANOREXIA NERVOSA

FIG 2. Daily activity counts during phases of anorexia nervosa. Nonbulimic (open circles)and bulimic (closedcircles) anorectics are compared foreach phase. Healthy control women are indicated by triangles.All subjects maintaineda stable weight (± 1.0 kg) during caloricmeasurement Subgroups are compared by two-tailed group I test, p <0.05.

heavier than nonbulimic anorectics, yet non-bulimic anorectics consumed significantlymore calories/day or per body surface area(Figure 1). Bulimic and nonbulimic anorecticshad similar daily activity counts (Figure 2).

The bulimic and nonbulimic anorecticsstudied after long-term weight recovery (Table3) were of similar weight. Although bulimicshad significantly more activity counts/daythan nonbulimic anorectics (Figure 2), bulimicanorectics at stable weight consumed fewercalories/day and per body surface area (Figure1) than nonbulimic anorectics.

For each phase of the study, bulimic andnonbulimic anorectics were compared to con-trols by a one-way analysis of covariance, ad-justing means of caloric intake/body surfacearea for motor activity. This analysis made noappreciable difference in terms of statisticalconclusions that bulimic and nonbulimic an-orectics had differences in energy-metabolismefficiency. Moreover, the same conclusionswere reached when caloric intake was adjustedfor body weight or for body mass index.

Previous investigators (1, 3) have noted sig-nificant differences in premorbid body weightbetween bulimic and nonbulimic anorectics.Thus we obtained data on premorbid weightfrom our subjects in order to determinewhether bulimic anorectics had greater pre-morbid body weight than nonbulimic anorec-tics. The two separate groups of weight-restored (short-term and long-term) anorecticswere combined and then separated by bulimicor nonbulimic behavior. Prior to weight loss,bulimic anorectics (n = 11) had been at amaximum of 104 ± 13% average weight (30).Nonbulimic anorectics (n = 9) had a trend (t= 1.96, p = 0.07, two-tailed t) towards a lowerpremorbid body weight, 94±9% average bodyweight. The two groups had been at similarlow body weight at some point in their illness,58 ± 6 versus 53 ± 9% average body weight.

Discussion

This study has shown that at several stagesof anorexia nervosa when weight remained

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

ENERGY METABOLISM IN ANOREXIA NERVOSA 441

stable, nonbulimic anorectics needed to con-sume a greater number of calories than bu-limic anorectics. Differences in activity did notaccount for these findings because bulimicshad similar or greater activity counts thannonbulimics. Differences in body weight alsodid not account for these findings. Bulimicanorectics, in fact, weighed more than non-bulimic anorectics at several stages of studyyet required a lower absolute caloric intake.Moreover, this difference in caloric intake be-tween bulimics and nonbulimics persistedwhen total daily caloric intake was expressedin terms of body surface area or body massindex.

It is unlikely that malabsorption was re-sponsible for increased caloric needs of thenonbulimics. Russell and Mezey (31) havedemonstrated that, during refeeding with ahigh-calorie liquid diet, anorectics were ableto absorb a normal proportion of ingested cal-ories. In addition, no patient in this study hadclinically observable edema. Thus, it is un-likely that weight gain from fluid accumulationcould account for the observed differences inbody weight or caloric intake.

The validity of this study relies upon theaccuracy of the estimate of caloric intake. In-vestigators have generally agreed that caloricintake can be estimated accurately by theweighing of all food consumed (32-34). Ourmethod of caloric estimation was similar andthe accuracy of our caloric determinations wasconfirmed by laboratory analysis (17).

Furthermore, we used precautions to pre-vent food from being binged and vomited orhidden and thrown away by restricting subjectsto a locked ward and by observing them atmeals and in the bathrooms. No patient onthe ward was allowed to have food other thanat meals. While some loss of food occurredoccasionally, it was rarely a significantproblem.

We continuously measured caloric intakeon underweight and short-term weight-restored anorectics for a mean of 11 and 23days, respectively. During this period of time,no subject gained or lost> 1.0 kg nor dem-onstrated any trend in weight gain or loss. Themean caloric intake during this period of timeshould be a reasonably accurate estimate ofenergy requirements for weight maintenance.Unfortunately, we were only able to measure

caloric intake in long-term weight-restoredanorectics for a mean of 4 days. While thisbrief evaluation period after long-term weightrestoration provided only limited data, themagnitude of the difference in caloric intakebetween bulimic and nonbulimic long-termweight-restored anorectics was similar to ourfindings in low-weight and short-term weight-restored anorectics.

Several investigators have observed that bu-limic anorectics have greater premorbid weightthan nonbulimic anorectics (1, 3). We havefound a similar trend. These premorbid weightdifferences might indicate that alterations inefficiency of energy metabolism predated theonset of bulimic or nonbulimic anorectic be-havior. However, it is also possible that yearsof chronic bingeing-and-purging behavior, orchronic starvation, might be responsible forthis alteration in energy-metabolism efficiency.

Other investigators (35, 36) did not cate-gorize anorectics into bulimic and nonbulimicsubgroups, but did find that previously obeseanorectics (presumably the bulimic subgroup)gained weight more rapidly, on the same foodintake, than anorectics who had previouslybeen of normal weight. Furthermore, StordyCt al (34) reported that previously obese ano-rectics, compared to those not previouslyobese, had a smaller increase in metabolic ratewith rising weight and a tendency to exhibita smaller thermic response to food. Thus,these data are also suggestive of differencesin energy-metabolism efficiency betweensubgroups of anorectics.

We were surprised to find another differencein caloric intake. The short-term weight-restored anorectics needed more calories (tomaintain a stable weight) than long-termweight-restored anorectics. [This finding isdiscussed at greater length elsewhere (37).]This was true for both bulimic and nonbulimicsubgroups. The reduction in caloric intake be-tween short- and long-term weight-restoredanorectics cannot be attributed to changes inactivity alone. It is not clear from this studywhether a reduction in caloric intake in in-dividual patients occurs in the months oftransition from short- to long-term weight re-covery because, with one exception, the short-term and long-term subjects were differentpeople. One subject (number 11) was studiedlongitudinally and did show a reduction in ca-

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

442 KAYE ET AL

loric intake between short-term and long-termweight restoration, but it is problematic todraw conclusions from one subject.

It is also possible that short-term and long-term weight-restored anorectics represent twodifferent pools of subjects: those with poor

outcome and those with good outcome, re-spectively. Increased caloric requirementmight contribute to poor outcome in the short-term weight-restored anorectics because itwould be difficult for these patients to eat suf-ficient calories to maintain weight. In contrast,the long-term weight-restored anorectics (thegood-outcome cohort) might have neededfewer calories at the time they were short-termrestored and thus have had an advantage inthe fight to maintain weight. A longitudinalstudy will be necessary to answer these ques-tions.

The findings reported in this paper are ofclinical relevance in advising anorectics aboutthe range of calories necessary to maintainweight after they finish a weight-restorationprogram. Nonbulimic anorectics would havegreater difficulty in maintaining their weight,both because of their lack of motivation to eatsufficient food and because of a less efficientenergy metabolism. Conversely, greater effi-ciency of energy metabolism in bulimic ano-rectics would tend to promote rapid weightgain and put bulimic anorectics at risk for be-coming obese. At present, only a minority ofanorectics are able to maintain a normal bodyweight permanently after weight restoration.A better understanding of caloric requirementsfor weight maintenance is necessary for thedevelopment of therapies that improve out-come in patients with anorexia nervosa.

References

1. Beaumont PVJ, George GCW, Smart DE. Dieters andvomiters and purgers in anorexia nervosa. PsycholMed l976;6:617-22.

2. Russell G. Bulimia nervosa an ominous variant ofanorexia nervosa. Psychol Med 1979;9:429-48.

3. Garfinkel PE, Moldofsky H, Garner DM. The het-erogeneity of anorexia nervosa. Arch Gen Psychiatry1980;37: 1030-40.

4. Casper RC, Eckert ED, Halmi KA, Goldberg SC,Davis JM. Bulimia its incidence and clinical impor-tance in patients with anorexia nervosa. Arch GenPsychiatry l980;37:1030-5.

5. Strober M, Salldon B, BurroughsJ, Morrell V. Validityof the bulimia-restricter distinction in anorexia ner-

vosa: parental personality characteristics and familypsychiatric morbidity. J Nerv Ment Dis l982;170:345-51.

6. Garner DM, Garfinkel PE, O’Shaughnessy M. Thevalidity of the distinction between bulimia with andwithout anorexia nervosa. Am J Psychiatry 1985;142:581-7.

7. Kaye WH, Ebert MH, Gwirtsman HE, Weiss SR. Dif-ferences in brain serotonergic metabolism betweennonbulimic and bulimic patients with anorexia ner-vosa. Am J Psychiatry 1984;14l:1598-601.

8. Rothwell NJ, Stock MJ. Luxuskonsumption, diet-induced thermogenesis, and brown fat:the case infavor.Clin Sd l983;64:19-23.

9. Hervey GR, Tobin G. Luxuskonsumption, diet-induced thermogenesis, and brown fat: a critical re-view. Clin Sci l983;64:7-l8.

10. Golay A, Schutz Y, Meyer HU, et al. Glucose-inducedthermogenesis in nondiabetic and diabetic obese sub-jects. Diabetes 1982;3l:1023-8.

11. Schutz Y, Bessard T, Jequier E. Diet-induced ther-mogenesis measured over a whole day. Am J ClinNutr 1984;40:542-52.

12. Segal KR, Gutin B. Thermic effects of food and ex-ercise in lean and obese women. Metabolism l983;32:581-9.

13. Morgan JB, York DA, Wasileska A, Portman J. Astudy of the thermic responses to a meal and to asympathomimetic drug (ephedrine) in relation to en-ergy balance in man. Br J Nutr 1982;47:2l-32.

14. American Psychiatric Association. Diagnostic andstatistical manual of mental disorders 3rd ed. Wash-ington, DC: APA, 1980.

15. Frisch RE. Food intake, fatness, and reproductiveability in anorexia nervosa. In: Vigersky R, ed. An-orexia nervosa. New York: Raven Press, 1977:149-61.

16. Falk JR. Halmi KA. Amenorrhea in anorexia nervosaexamination of the critical body weight hypothesis.Biol Psychiatry 1982;l7:799-806.

17. Petersen R, Kaye WH, Gwirtsman HE. Estimationof caloric intake for patients hospitalized for eatingdisorder. J Am Diet Assoc l986;86:490-2.

18. Khosla T, Billewicz WZ. Measurement of change inbody weight Br J Nutr 1964;18:227-39.

19. Robinson MF, Watson PE. Day-to-day variations inbody weight of young women. Br J Nutr 1965;19:225-35.

20. Pomno LI, Rapoport JL, Behar D, Sceery W, IsmondDR. Bunney WE. A naturalistic assessment of the motoractivity of hyperactiveboys. IComparison with normalcontrols. Arch Gen Psychiatry l983;4068l-7.

21. Wehr TA, Goodwin FK. Biological rhythms and psy-chiatry. In: Arieta 5, Brodie HK, eds. Americanhandbook of psychiatry. New York: Basic Books,1981:46-74.

22. Colburn T, Smith BM, Guarini J. An ambulatory ac-tivity monitor with solid-state memory. InstrumentSociety of America Transactions 1976;15:l49-54.

23. Webster AJF. The energetic efficiency of metabolism.Proc Nutr Soc 198l;40:l21-8.

24. Bastow MD. Anthropometrics revisited.Proc NutrSoc 1982;4l:381-8.

25. Keys A, Fidanza F, Karvonen MJ, Kimura N, Taylor

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from

ENERGY METABOLISM IN ANOREXIA NERVOSA 443

HL. Indices of relative weight and obesity. J ChronicDis 197225:329-43.

26. Cunningham JJ. Body composition and resting met-abolic rate: the myth of feminine metabolism. Am JClin Nutr l982;36:72l-6.

27. Durnin JVGA. Body weight, body fat and the activityfactor in energy balance. In: Apfelbaum M, ed. Energybalance in man. Paris, France: Masson et cie, 1973:141-50.

28. Dubois D, Dubois EF. Clinical calorimetry. A formulato estimate the approximate surface area if height andweight be known. Arch Intern Med 1916;l7:863-71.

29. Department of Biomathematics. BMDP statisticalsoftware. Los Angeles, CA: UCLA Press, 198 1:359-87.

30. Metropolitan Life Insurance Company. New weightstandards for men and women. (StatisticalBulletin,New York, NY no 40) Metropolitan Life Insurance

Company, 1959;l-ll.

31. Russell GFM, Mezey AG. An analysis of weight gain

in patients with anorexia nervosa treated with high-calorie diets. Clin Sci 196223:449-61.

32. Marr JW. Individual dietary surveys: purposes andmethods. World Rev Nutr Diet 197 l;13:l05-64.

33. Pekkarinen M. Methodology in the collection of food-consumption data. World Rev Nutr Diet l970;12:145-71.

34. Whiting MG, Leverton RM. Reliability of dietary ap-praisal: comparisons between laboratory analysis andcalculation from tables of food values. Am J PublicHealth 1960;50:8l5-23.

35. Stordy BJ, Marks V, Kalucy RS, Crisp All. Weightgain, thermic effects of glucose, and resting metabolicrate during recovery from anorexia nervosa. Am JClin Nutr 1977;30:138-46.

36. Walker J, Roberts SL, Halmi KA, Goldberg SC. Ca-loric requirements for weight gain in anorexia nervosa.Am J Clin Nutr l979;32: 1396-400.

37. Kaye WH, Gwirtsman H, George T, Ebert MH, Pe-terson R. Caloric consumption and activity levels afterweight recovery in anorexia nervosa. a prolonged delayin normalization. Int J Eat Dis 1986;5:489-502.

at University of P

ittsburgh HS

LS on M

arch 6, 2007 w

ww

.ajcn.orgD

ownloaded from