Secretin as a Treatment for Autism: A Review of the Evidence

Barbara E. Esch1,2 and James E. Carr1

Secretin is used in the United States for diagnosis of pancreatic gastrointestinal (GI) dys-function and disease. Repeated therapeutic use has not been approved. Widespread interest

in secretin as a treatment for autism followed media reports of behavioral improvements inan autistic child who received the hormone during a GI diagnostic procedure. Internationaldemand for secretin soared in the absence of experimental evidence of its efficacy for aut-

ism. This review presents a brief history of secretin’s rise to popularity and summarizesresearch on secretin as a treatment for autism. Seventeen studies are reviewed comparingthe effects of secretin forms, dosage levels, and dosing intervals on outcome measures with

approximately 600 children. Twelve of 13 placebo-controlled studies failed to demonstratethe differential efficacy of secretin. Implications for advocating treatment in the absence ofempirical evidence are discussed.

KEY WORDS: Secretin; autism; efficacy.

INTRODUCTION

Autism is a severe, lifelong, neurodevelopmen-tal disorder characterized by impairments in multi-ple domains. It is usually identified before the ageof three by behavioral assessment of qualitativeimpairment in social interaction, qualitative impair-ment in communication, and markedly restricted,repetitive patterns of behavior and interests (Amer-ican Psychiatric Association, 1994; APA). Preva-lence has been estimated at 2–5 per 10,000 (APA,1994) although some estimates are as high as 1 in250 (Bertrand et al., 2001). Approximately 75% ofchildren with autism also meet the diagnostic crite-ria for mental retardation and about half neveracquire functional speech (Hardman, Drew, &Egan, 1999). The etiology of autism remainsunknown; however, recent studies suggest a likely

genetic basis for the disorder (see Veenstra-Vander-weele & Cook, 2003).

Although there is no known cure for autism, anumber of intervention strategies have been evalu-ated over the years (Herbert, Sharp, & Gaudiano,2002), with extremely promising results in some areas(e.g., applied behavior analysis; Smith, 1999). A fewexamples of the varied and numerous treatments rec-ommended for autism include vitamin supplements,dietary manipulations, auditory integration training,facilitated communication, hyperbaric oxygen treat-ment, live- and stem-cell therapy, anti-fungal treat-ment, detoxification for heavy metals, biofeedback,and craniosacral therapy.1 Unfortunately, many ofthese treatments have not been adequately testedusing proper scientific methods.

One treatment for autism that has beenrecently popularized is secretin. Secretin is a hor-mone secreted by the duodenum in response toincreased acidity in the stomach. It stimulates therelease of bicarbonate and enzymes from the pan-1 Western Michigan University, Kalamazoo, MI.

2 Correspondence should be addressed to Barbara E. Esch,

Department of Psychology, Western Michigan University,

1903 W. Michigan Avenue, Kalamazoo, MI 49008-5439;

e-mail: barbesch @gate.net

1For commentary on these treatments, see

www.acsh.org/publications/priorities/1301/autism.html and

www.healing-arts.org/children.

Journal of Autism and Developmental Disorders, Vol. 34, No. 5, October 2004 (� 2004)

5430162-3257/04/1000-0543/0 � 2004 Springer Science+Business Media, Inc.

creas, thus acting as ‘‘a sort of built-in Alka-Seltzerdispenser’’ (Shutt, 1998, p. 21). Porcine secretin wasapproved by the United States Food and DrugAdministration in 1981 for use in the diagnosis ofgastrointestinal (GI) disorders. In the United States,secretin is commonly given in a single dose topatients experiencing pancreatic insufficiency duringa diagnostic procedure for this GI complaint or toaid in the diagnosis of pancreatic tumor. A syn-thetic form of human secretin is available which hasbeen found to be equipotent in its pharmacologicaleffects to the porcine form (Christ et al., 1988).2

In 1998, Horvath et al. reported a case series inwhich three children with autistic spectrum disorderreceived a single infusion of intravenous porcinesecretin during diagnostic GI endoscopy for chronicdiarrhea. Within 5 weeks, all children evidenced notonly amelioration of their GI symptoms but parentsalso reported dramatic improvement in their chil-dren’s communication and social behavior.3 Theauthors speculated that these clinical observationsmight suggest an association4 between GI and brainfunction in individuals with autism although themechanism for this link was not (and has not been)clearly defined. Horvath (2000) noted that empiricalstudies, not case reports, would be necessary to fullyarticulate the role of secretin in central nervous sys-tem function and emphasized the importance of pla-cebo control in such future studies. By contrast,improvements that were seen in the 1998 case serieswere incidental to the endoscopic medical procedurefor GI complaints; thus, no control group wasincluded nor were behaviors that were specificallyrelated to autism assessed.5 Thus, althoughimprovements may have occurred, the methods

inherent in this type of study preclude attributingany such improvements to the action of secretin.

On October 7, 1998, NBC’s Dateline reportedthe dramatic improvement observed in one of theHorvath et al. (1998) participants. Widespread In-ternet and lay media attention ensued (Coniglioet al., 2001; Sandler et al., 1999) and parents imme-diately began to request secretin infusions for theirchildren with autism (Aman & Armstrong, 2000),often paying highly inflated prices for the injectionswhich at times reportedly contained no secretin inthe preparation (National Institutes of Health NewsAlert, 1999). The overwhelming demand for secretinjeopardized the nation’s available supply (Rimland,1999). Websites for The Autism Research Institutein San Diego (www.autism.com/ari) and the Univer-sity of Sunderland’s Autism Research Unit in theUnited Kingdom (osiris.sunderland.ac.uk/autism)began to post regular research updates and anec-dotal reports of the possible benefits of secretin aswell as provide commentary on secretin-relatedissues. Sandler et al. (1999) and Owley et al. (1999)speculated that thousands of children might havereceived secretin injections following the initialhighly publicized case, all in the absence of empiri-cal support.

Accompanying the early media and Internetcoverage of secretin were many voices urging cau-tion to limit or suspend its therapeutic use until effi-cacy studies6 could be completed. Commenting onthe downside of the sudden media attention tosecretin, Volkmar (1999) stated, ‘‘what makes aninteresting television program may not, of course,be the same as what makes good science’’ (p. 1844).He admonished physicians to help patients and theirfamilies make informed decisions, presumably basedon empirical data instead of unfounded claims.Reacting to the interest of some parents in contin-uing to seek unproven secretin treatment for theirchildren, Volkmar (2000) stated that, unfortunately,treatment efficacy could not be judged by treatmentpopularity.

Although McMillin, Richards, Mein, andNelson (1999) suggested that it was entirely plausi-

2The chemical structure of these two forms of secretin differs by

two amino acids (cf. Carey et al., 2002), a subtle but potentially

important difference although the direction or relevance of any

difference (e.g., clinical, biologic) has not been determined.3Horvath (2000) noted that in the majority of patients in a simi-

lar study (Horvath, Papadimitriou, Rabsztyn, Drachenberg, &

Tildon, 1999), gastrointestinal changes were noted after a single

dose of secretin whereas behavioral improvements were seen

gradually and after repeated injections.4GI dysfunction in many children with autism has led to common

speculation of a possible causal connection between abnormal

brain function and intestinal tract problems. This connection is

often referred to as the ‘‘brain-gut’’ hypothesis. However, cau-

sality in either direction has not been substantiated and thus the

‘‘hypothesis’’ remains as such.5The authors reported that developmental and psychological eval-

uations were administered prior to the procedure.

6Some of the important variables to include in these studies were

identified by Aman and Armstrong (2000). In a survey of par-

ents whose children had previously received secretin, they que-

ried dosage level, patient demographics, presence of GI

disturbances prior to treatment, latency and duration of treat-

ment effect, and any observed side effects. All of these compo-

nents were ultimately addressed in the (collective) subsequent

studies reviewed here.

544 Esch and Carr

ble that neurologic syndromes (e.g., autism) that arefrequently accompanied by GI symptoms could belinked to an ‘‘abdominal nervous system,’’ Shutt(1998) warned of the ‘‘. . . dangers in tinkering withthe ‘bottle of hydrochloric acid’ we all carry in ourstomachs’’ (p. 24). He suggested some childrenmight form immunities to the porcine form of secre-tin, resulting in rejection of their own secretin withpotentially devastating effects.

DiCicco-Bloom (1998) advised parents not toallow secretin infusion without evidence of its effi-cacy in controlled studies that replicated andextended the incidental findings of the Horvathet al. (1998) case series. In remarks addressed toparents, he contrasted the available evidence, char-acterized as ‘‘merely a report’’ (p. 24), with specificresearch procedures that would provide databasedinformation. He reminded parents that the uncon-trolled case series report (Horvath et al.) had notyet proven the efficacy of secretin as a treatment forautism. Other concerns were the off-label (i.e., non-FDA-approved) use of secretin in the absence ofcontrolled scientific studies (American Academy ofChild and Adolescent Psychiatry Policy Statement,1999) and the possibility of allergic reaction if mul-tiple doses of the animal-derived form of secretinwere infused (Hirsch, 1999).

Given the remarkable claims regarding secre-tin’s effects, the overt significance of treating autismvia intravenous hormone therapy, and the flurry ofrecent scientific attention secretin has received, thepurpose of the current paper is to critically reviewthe peer-reviewed quantitative research on secretinas a treatment for the symptoms of autism. Articleswere identified through the PsycINFO� and MED-LINE� databases from 1986 to 2002 using the key-words ‘‘autism’’ and ‘‘secretin.’’ An earlier reviewby McQueen and Heck (2002) reported the resultsof eight7 efficacy studies published through 2001.The purpose of the present paper is to expand theearlier review to include additional studies that wereeither omitted or have been subsequently published.

Seventeen studies were identified and includedin our review (see Table I). Sixteen studies includedchildren as participants and one study evaluatedsecretin with adults. All participants had diagnoseswithin the autism spectrum disorder (autistic disor-der, pervasive developmental disorder [PDD], orpervasive developmental disorder not otherwise

specified [PDDNOS]). Because the initial case seriesreport (Horvath et al., 1998) leading to this line ofresearch suggested a possible relation between aut-ism and GI disorder, some of the subsequent studiesreviewed here either specifically included orexcluded that demographic characteristic. The stud-ies ranged in duration from 3 to 16 weeks. Thirteenof the studies were double-blind and placebo-con-trolled and four were uncontrolled open-label casestudies. Porcine secretin was used in 11 of the inves-tigations, synthetic human secretin was administeredin 4, one used a homeopathic secretin preparation,and another compared both the biologic and syn-thetic forms of porcine secretin to placebo. A singledose of secretin was used in 14 studies; the remain-ing three studies investigated the effects of multipledoses.

RESEARCH COMPARING SYNTHETIC

SECRETIN TO PLACEBO

In the first controlled study8 of the effects ofsecretin to ameliorate the symptoms of autism,Sandler et al. (1999) studied 56 children with a diag-nosis of autism or PDD. This double-blind, placebo-controlled study used a single dose of humansynthetic secretin. Randomly assigned participantsreceived either a single infusion (slow injection) ofsecretin or a single infusion of a volume-comparablesaline placebo.

Autism-related behaviors were rated by par-ents using the Autism Behavior Checklist (AuBC)and the Clinical Global Impression Scale (CGIS).Study clinicians and teachers rated communicationbehaviors using the Vineland Adaptive BehaviorScales (VABS) and clinicians assessed adversetreatment effects with the Treatment EmergentSymptoms Scale. Measurements were taken2 weeks and 1 week before treatment, at 1 and2 days post-infusion, and at the end of weeks 1, 2,and 4 post-treatment. Although both groups dem-onstrated improvement (i.e., decrease in severity ofsymptoms) on 6 out of 16 outcome measures,results failed to show a significant differencebetween the secretin and placebo groups on any ofthe outcome measures. No particular subgroup ofresponders was identified with respect to age,severity of autism symptoms, GI symptoms, or the

7The review also included two studies on the safety of secretin

and its intestinal permeability.

8Sponsored by the National Institute of Child Health and Human

Development (NICHD).

545Secretin as a Treatment

Table

I.Summary

ofControlled

StudiesandCase

Reports

Outcomes

Article

Typeofstudy

N

Participant

ages

Diagnoses

Typeand

Dose

aClinician

rating

Caregiver

rating

Other

Positive

results

Careyet

al.(2002)

Randomized,double-blind,

placebocontrolled,crossover

82–8yr

AD

orPDD

SomewithGI

Single

dose

2CU/kgcompounded

Synthetic

secretin

��

No

Chez

etal.

(2000)—

study1

Uncontrolled,open-label

56

Mean6.4

yr

AD

orPDDNOS

SomewithGI

Single

dose

2IU

/kgporcinesecretin

��

Yes

TransientO1wk

Chez

etal.

(2000)—

study2

Randomized,double-blind,

placebocontrolled,crossover

25

Mean6.0

yr

AD

orPDDNOS

SomewithGI

Single

dose

2IU

/kgporcinesecretin

�No

Coniglioet

al.(2001)Randomized,double–blind,

placebocontrolled

60

3–10yr

AD

Single

dose

2CU/kgporcinesecretin

��

3wks–yes

6wks–no

Corbettet

al.(2001)

Randomized,double-blind,

placebocontrolled,crossover

12

4–12yr

AD

Excluded

GI

Single

dose

2IU

/kgporcinesecretin

��

No

Dunn-G

eier

etal.(2000)

Randomized,double-blind,

placebocontrolled

95

2–7yr

AD

Single

dose

2CU/kgporcinesecretin

��

�No

Horvath

etal.(1998)Case-seriesreport

33–5yr

AD

orPDDNOS

AllwithGI

Single

dose

2IU

/kg

porcineb

secretin

��

�Yes

Kernet

al.(2002)

Randomized

double–blind

placebocontrolled

crossover

19

3–10yr

AD

orPDD/

PDDNOS

SomewithGI

Single

dose

2CU/kgporcinesecretin

��

�Yes

(GIsubgroup

only)

Lightdale

etal.

(2001)

Uncontrolled,open-label,

blindrating

20

3–8yr

AD

AllwithGI

Single

dose

3CU/kgporcinesecretin

��

No

Molloyet

al.(2002)

Randomized,double-blind,

placebocontrolled,crossover

42

2–15yr

AD

SomeGIexcluded

Single

dose

2IU

/kgSynthetic

human

secretin

��

�No

Owleyet

al.(1999)

Randomized,double-blind,

placebocontrolled,crossover

20

3–12yr

AD

Single

dose

2CU/kgporcinesecretin

��

No

Owleyet

al.(2001)

Randomized,double-blind,

placebocontrolled,crossover

56

3–12yr

AD

Single

dose

2CU/kgporcinesecretin

��

No

Roberts

etal.(2001)

Randomized,double-blind,

placebocontrolled

64

2–7yr

AD

Multiple

doses(2)

2ml/kgporcinesecretin

��

�No

Robinson(2001)

Uncontrolled,open-label

12

24–43yr

AD

Multiple

doses

homeopathic

secretin

c

Orallytw

icedailyfor

12wks

�No

546 Esch and Carr

use of adjunctive medication. No treatment-limitingside effects were observed in any of the children.

In discussing the suspected placebo effects fromtheir 1999 study (i.e., 30% of both groups showedimprovement on outcome measures), Sandler andBodfish (2000) recalled that a majority of parents inthe 1999 study continued to express interest insecretin even after being informed of their child’sgroup assignment and in the absence of clinicaleffect attributable to secretin. The authors offeredseveral factors that could account for this. Mostparents have a strong desire to improve their child’squality of life. This, coupled with possible dissatis-faction with other treatment options, may contrib-ute to continued interest in a treatment that has notbeen conclusively disproved. Moreover, the disrup-tions inherent in living with a child who has autismmight give rise to seeking out any reasonable treat-ment that might promise relief, despite the lack ofscientific support.

There were some limitations of the Sandleret al. (1999) study. It was short-term; a few weeksmay not have been enough time to observe signifi-cant behavior change. In addition, multiple dosesof secretin may have demonstrated greater effectsthan a single dose. However, the Horvath et al.(1998) case series also used a single dose andimprovements were anecdotally reported within asimilar time frame (i.e., 4–5 weeks). Another limi-tation concerns the inclusion criteria used in thestudy. Specific subgroups of participants with dif-fering potentials for response to treatment mightbe identified using certain diagnostic instruments(e.g., children with particular GI problems mightbe more responsive to secretin treatment than chil-dren without these symptoms). Finally, whereasmost anecdotal reports of positive secretin effectsfollowed infusion of the porcine derivative ofsecretin, this study used synthetic human secretin.These limitations notwithstanding, Alexander(1999) recommended that, based on the Sandleret al. results, the use of secretin as a treatment forautism be suspended until further empirical studiescould be completed.

Carey et al. (2002) also evaluated the effectsof synthetic human secretin. In a randomized, dou-ble-blind, placebo-controlled study, eight childrenreceived first either secretin or placebo, and then‘‘crossed over’’ (via a crossover design) to thealternative treatment 4 weeks later. Thus, partici-pants served as their own controls. Changes in aut-ism-related abnormal behaviors were evaluatedS

andleret

al.(1999)

Randomized,double-blind,

placebocontrolled

56

3–14yr

AD

orPDDNOS

SomeGIexcluded

Single

dose

0.4

mcg/kgsynthetic

human

secretin

��

No

Sponheim

etal.(2002)

Randomized,double-blind,

placebocontrolled,crossover

63–12yr

AD

NopronouncedGI

Multiple

doses(3)

4CU/kgsynthetic

human

Secretin

��

�Smalldifferences

(noclinical

significance)

Uniset

al.(2002)

Randomized,double-blind,

placebocontrolled

85

3–12yr

AD

orPDDNOS

Single

dose

2CU/kgporcinesecretin

or

0.4

mcg/kgSynthetic

porcinesecretin

��

No

aSecretinpotency

isdesignatedin

clinicalunits(C

U)orinternationalunits(IU);conversioncomparisonis0.2

mcg

=1CU.

bPersonalcommunicationwithfirstauthor,January

8,2003.

cSecretin6C

(centesimaldilution).

Note:AD

=autistic

disorder,GI=

gastrointestinalcomplaint,

PDD=

pervasivedevelopmentaldisorder,PDDNOS=

pervasivedevelopmentaldisorder

nototherwisespecified.

Thedescriptor‘‘uncontrolled’’refers

tonoplacebocontrolandnonrandomized

samples.

547Secretin as a Treatment

using the Aberrant Behavior Checklist (ABC).Consistent with the Sandler et al. (1999) investiga-tion, Carey et al. found no reliable improvementsattributable to secretin. The study was treated as aclinical replication series due to small sample sizeand data were analyzed accordingly. The studywas further limited by the use of only one out-come measure. It is possible that clinically signifi-cant change might have been detected with the useof additional measures. Similarly, a single dose ofsecretin may have been insufficient to effect mea-surable behavior change. Finally, the study mayhave been limited by carry-over effects due to thecrossover design. Data are not definitive on theduration of possible secretin effects so the possibil-ity of treatment-order effects within participantscould not be eliminated.

In a study that addressed some of the limita-tions of the Carey et al. (2002) investigation,Molloy et al. (2002) reported the results ofanother randomized, double-blind, placebo-con-trolled crossover study of 42 children using a sin-gle dose of synthetic human secretin in which acomprehensive battery of tests was used to assessbehavioral outcomes. Measurements were takenfive times (prior to the first infusion, at week 3,at week 6 prior to the crossover infusion, and atweeks 9 and 12). Receptive and expressive lan-guage was evaluated by two speech and languagepathologists using the Peabody Picture VocabularyTest-3 (PPVT-3) and the Receptive LanguageScale of the Mullen Scales of Early Learning(MSEL-RLS). Cognitive skills and autism-relatedbehaviors were assessed by a clinical psychologistusing selected tests of the Merrill–Palmer Scale,the Developmental Test of Visual Perception(DTVP), the Childhood Autism Rating Scale(CARS), and the Gilliam Autism Rating Scale(GARS). In addition, parents were asked to peri-odically provide information about their child’smedical condition and stool patterns. Blood-chem-istry profiles were also obtained for all partici-pants.

No significant differences were found betweenthose who received secretin compared to placebo,nor were there treatment-order effects betweengroups who received either secretin or placebo asthe first treatment. Negative results were alsoobtained for differential effects of secretin on a sub-group of participants with GI symptoms. The studymay have been limited because of the self-referredpopulation with males being heavily over-repre-

sented; however, the use of random assignmentshould mitigate these concerns.

In a two-part crossover study, and the second(cf. Sandler et al., 1999) secretin investigation spon-sored by NICHD, Owley et al. (1999) studied 20children with a diagnosis of autism using a random-ized, double-blind, placebo-controlled crossover trialof porcine secretin. Participants received eithersecretin or a saline placebo at baseline followed bythe other substance at the second infusion (week 4).The primary outcome measure was the AutismDiagnostic Observation Scale-Generic (ADOS-G),which provided scores in social communicationbehavior. Additional behavioral outcome data wereobtained at the same intervals (baseline, after weeks4 and 8) using the DTVP or the fine motor sub-scale of the Mullen Scales of Early Learning, thePPVT or the MSEL-RLS for receptive languageassessment, and the CGIS. At the same intervals,parents were interviewed using the VABS. They alsocompleted the GARS and the ABC, Communityversion (ABC-C) at baseline and at weeks 2, 4, 6,and 8.

Although randomly assigned, the two groups(secretin-first and placebo-first) differed at baselineon the ADOS-G social impairment score with thesecretin-first group functioning at a significantlylower level. Conversely, baseline scores on theADOS-G stereotypy measure were higher for thesecretin-first group. Owley et al. (1999) speculatedthat the secretin-first participants might represent agroup that is less likely to benefit from secretin infu-sion, although no ceiling effect was exhibited by thisgroup in which improvement would be difficult todetect.

The placebo–secretin group did not improveafter the second infusion (secretin). This is a rele-vant finding because the group acts as its own con-trol and no carry-over effects are expected followingthe placebo infusion. Although minimal improve-ment was noted in both groups during the first4 weeks, there were no statistically significant differ-ences between the two groups at the end of 8 weeks.The initial improvement in both groups could beattributed to expectancy effects since the trend forboth groups was the same. None of the participantsdemonstrated adverse effects that appeared relatedto secretin infusion.

The Owley et al. (1999) study was part of a lar-ger multi-site investigation (Owley, McMahon, &Cook, 2001) and these preliminary data werereported in an effort to inform parents about the

548 Esch and Carr

potential efficacy of secretin as a treatment for aut-ism. As such, the small sample size is recognized asa temporary limitation. Of more concern is the pos-sibility that positive effects might have been obtainedwith a stronger dose or with repeated infusions.However, the dosage strength was the same as thatof Horvath et al. (1998) in which improvement wasreported following only one infusion.

Owley et al. (2001) subsequently reported theresults of the multi-site investigation using a ran-domized, double-blind, placebo-controlled, cross-over design to study the effects of secretin in 56children with autism. The outcome measures andtesting intervals were identical to those in the preli-minary study (i.e., Owley et al., 1999) and resultsagain failed to demonstrate any significant effectsattributable to secretin.

Dunn-Geier et al. (2000) used a randomized,double-blind, placebo-controlled trial to examinethe effects of a single dose of porcine secretin in 95children with autism. Clinicians and parents ratedsocial interaction, communication, and behavioralcharacteristics before secretin infusion and at a 3-week follow-up using the Preschool LanguageScale-3 (PLS-3), the CARS, the AuBC, as well asGI and side-effect questionnaires. No seriousadverse effects were reported nor were any signifi-cant problems noted in terms of safety. Resultswere consistent with previous findings. The studyfailed to demonstrate any differential treatmenteffect from a single infusion of secretin and thuscould not corroborate the earlier report (Horvathet al., 1998) of improvement in language andbehavior of children with autism following secretinadministration.

The anecdotal success of secretin prompted atwo-part clinical investigation by Chez et al.(2000). The first study employed an uncontrolledopen-label trial of porcine secretin with 56 childrendiagnosed with either autism or PDDNOS. Behav-ioral effects were measured with the CARS, whichwas completed by parents at baseline and again atfollow-up approximately 4 weeks later. Treatmenteffect was analyzed both statistically and clinically.Clinically significant change was defined byresearchers as a 6-point or greater improvement onthe CARS.

In terms of clinical improvement, 13 partici-pants (23%) classified with severe autism demon-strated the requisite gain for significance, althoughno dramatic improvements in speech or typicalautistic characteristics were observed. The majority

had either clinically insignificant change, no change,or were rated as ‘‘worse.’’ Some of the adverseeffects noted by parents included increasedhyperactivity, increased agitation, and decreasedfocusing and responsiveness to others. Some parentsreported improvements in GI function, eye-contact,and expressive and receptive speech. However, theseimprovements dissipated within a week. There wasa statistically significant difference from baseline tofollow-up for the entire group in several categoriesof the CARS including relating to people, emotionalresponse, and verbal communication. While statisti-cally significant, this difference did not qualify asclinically significant according to the criterion of aP6-point improvement on the CARS.

To determine whether the statistically signifi-cant improvements were due to rater bias or trueeffects of the hormone, a second study was con-ducted. In order to maximize the probability ofdetecting a difference in this follow-up study, arandomized, double-blind, placebo-controlled,crossover design was used. Participants were alter-nately assigned to one of two groups upon entryand received either porcine secretin or placebo atbaseline, followed by the alternate injection at the4-week follow-up. Some of the participants whowere reported by their parents as ‘‘improved’’ inthe first study were selected for Study 2. Chezet al. (2000) reasoned that if improvements weredue to the secretin infusions, then a repeat injec-tion should make previously subtle changes moredetectable.

A total of 25 children completed Study 2.Assessments (neurologic interviews, clinical assess-ments, and CARS) were completed at baseline andat the end of 4 and 8 weeks. In addition, parentskept diaries of any behavior changes they observedin their children. The results from Study 2 failed todemonstrate any significant difference in CARSscores between the two groups. Even though all rat-ers were blinded to eliminate bias, the resultsshowed that parents reported (the unknown) salineplacebo as beneficial as secretin. This finding, cou-pled with the use of a crossover design, argues foran expectancy effect and against any clinicalimprovement provided by secretin. The interpreta-tion of reporting bias is further supported by thetransient gains observed in Study 1 because theblind Study 2 failed to demonstrate even mildimprovements in behavior change.

Several criticisms of the Chez et al. (2000) stud-ies were made by Rimland (2000); two issues in

549Secretin as a Treatment

particular should be mentioned. The lack of a wash-out period to eliminate residual effects of secretinwas cited by Rimland as the ‘‘most severe problem’’(p. 95) limiting the findings of the second study. Ifsuch carry-over effects of a drug are present, subse-quent treatment effects (or lack of effects) may bemasked. Thus, conclusions drawn from such dataare necessarily circumscribed. However, Chez andBuchanan (2000) dismissed the Rimland charge,stating that all participants were drug-free for a per-iod of 6–8 weeks prior to receiving injections, a per-iod that was more than adequate for effects todissipate considering secretin has a drug half-life of2 minutes when administered intravenously.

The second criticism concerned sample selec-tion. At the time of the first study, approximately80% of the participants were receiving medications,a statistic that Rimland (2000) asserted rendered thesample atypical. However, Chez and Buchanan(2000) defended their sample as representative ofthose reported in the literature with respect to varieddrug regimens. Moreover, in other studies reviewedin this article that also reported on-going pharmaco-therapy with their participants (Dunn-Geier et al.,2000; Owley et al., 1999, 2001; Sandler et al., 1999),no evidence was found for the differential efficacy ofsecretin.

Corbett et al. (2001) also used a randomized,double-blind, placebo-controlled crossover design toexamine the effects of porcine secretin in 12 childrenwith autism. Outcome measures taken at baselineand 1 week post-treatment included the ADOS-Gto assess social and communication improvement,the Minnesota Preschool Affect Rating Scales(MNPARS) to assess change in affect, and theCommunication and Symbolic Behavior Scales(CSBS) to determine improvement in expressive lan-guage skills. Neurologic evaluations were also com-pleted at each visit, and parents kept a daily log ofGI-related episodes.

In general, the results did not demonstrate theefficacy of secretin over placebo. Although twovariables on the MNPARS (i.e., affect and activity)achieved statistical significance, the authors did notinterpret these results since they occurred in isola-tion (i.e., the CSBS showed no difference in affectbetween groups). None of the other dependent mea-sures showed significant differences on behavioral,social, or communication measures. The study mayhave been limited by the testing schedule, whichwas restricted to a single administration 1 weekafter treatment; however, other studies (e.g., Molloy

et al., 2002; Sandler et al., 1999) in which repeatedtesting was done after treatment also failed to dem-onstrate the efficacy of secretin.

Coniglio et al. (2001) conducted a randomized,double-blind, placebo-controlled trial of secretinwith 60 children with autism who received a singledose of either porcine secretin or a saline placebo.Behaviors were assessed at baseline and at 3 and6 weeks following injection using the CARS,GARS, and the PLS-3. None of these measuresshowed significant differences at 6 weeks betweenchildren receiving secretin and those administered aplacebo. There was marginally statistically signifi-cant improvement in autistic behaviors in the treat-ment group after 3 weeks; however, the authorswere unable to identify the characteristics of a sub-group (e.g., GI responders) responsible for thisdifference.

Unis et al. (2002) recently reported the resultsof a randomized, double-blind, placebo-controlledinvestigation that compared two forms of secretin(synthetic porcine and biologic porcine) to placeboin addition to evaluating the benefit of secretin forchildren with GI problems. Eighty-five children withautism were evaluated for changes in language skills(Expressive One-Word Picture Vocabulary Test-Revised and MacArthur Communicative Develop-ment Inventory [MCDI]), social and communicationskills (ADOS-G and the Secretin Outcome Survey),and problem behaviors (ABC-C).

A comprehensive analysis of the results failedto show evidence for the superiority of either bio-logic porcine secretin or the synthetic porcine formover placebo in reducing the symptoms of autism.Similarly, no evidence of secretin efficacy wasfound for the subgroup of responders with GIproblems. In emphasizing the importance ofincluding multiple, and qualitatively different, out-come measures (i.e., direct observation and ratingscales) in such investigations, Unis et al. (2002)noted the comparatively high expectancy effects inthis and other studies (Coniglio et al., 2001; Dunn-Geier et al., 2000) that incorporated rating-scalemeasures. It is important to note that, by contrast,direct observation tends to yield smaller expectancyeffects9 and thus would mitigate (to some extent)overestimations.

9While multiple outcome measures (i.e., objective and subjective)

may be potentially problematic, this is less a concern when

results of these measures are concordant as they are in the

majority of studies reviewed here.

550 Esch and Carr

RESEARCH EVALUATING MULTIPLE DOSES

OF SECRETIN

The first study to investigate the effects ofrepeated doses of secretin was reported by Robertset al. (2001) in which 64 children with autism wererandomly assigned to receive two doses of eithersecretin or a placebo, 6 weeks apart. This double-blind study used a variety of outcome measures(ADOS-G, Leiter International Performance Scale-Revised or existing VABS data, PLS-3, visual per-formance tasks, AuBC, a GI questionnaire, and aside-effect rating scale) to rate improvements in typ-ical autistic behaviors and language and/or cogni-tive functioning. With the exception of the cognitivetests (completed only at baseline and follow-up), allassessments were repeated 3 weeks after each injec-tion.

No significant treatment differences were foundbetween the secretin and placebo groups. A furtheranalysis of participants by subgroups (i.e., GI symp-tomatology, cognitive level, and history of regres-sion) also failed to reveal any differential effect ofthe drug. Although parents reported markedimprovement in some cases, these gains were notclinically sustained. Moreover, the fact that suchimprovements were noted in both the treatment andplacebo groups strongly argues for explanatory vari-ables such as maturation and practice and expec-tancy effects and against the differential action ofsecretin.

In the only study with adult participants,Robinson (2001) reported an uncontrolled, open-label, clinical pilot investigation of the effects ofmultiple doses of homeopathic secretin. Twelveadults with a diagnosis of autism were evaluated bycaregivers at a residential care facility over 14 weeks(2 weeks in baseline, 12 weeks in treatment). Thesecretin preparation was administered orally twicedaily in a glass of water. Behaviors were measuredonce each week using the CARS.10

Comparisons of mean scores pre-treatment andweekly during treatment showed a surprisingincreasing trend, suggesting a worsening of symp-toms while on secretin, but these differences failedto reach statistical significance. Thus, althoughsecretin was not shown to significantly increaseautistic symptoms, neither was it found to decreasethose symptoms.

As an uncontrolled, open-label study, thisresearch is limited in its utility toward extendingour understanding of any relation between secretinand autism. Reliability is impacted due to inconsis-tency in raters across the duration of the study andthe absence of inter-rater reliability assessment. Inaddition, rater bias is possible because of theunblind design. Further, this is not a placebo-con-trolled study, which limits any conclusions aboutdrug effects. Finally, the homeopathic preparationis not described with respect to its equipotence toeither the porcine or synthetic form of secretin and,therefore, comparisons with other studies arerestricted.

Thus, while this study is unique in its partici-pant sample (adult), form of secretin (homeopathic),and frequency of dose (twice daily for 12 weeks),the results must be interpreted with considerablecaution due to its methodological limitations. Nev-ertheless, it should be noted that the study’s out-come is similar (i.e., no identifiable secretin effect)to most of the other studies reviewed in this article.

Sponheim, Oftedal, and Helverschou (2002)also investigated the effects of repeated doses ofsecretin. In this randomized, double-blind, placebo-controlled crossover design, six children receivedthree injections of secretin and three of placebo,randomized, every 4 weeks for a total of six injec-tions. It should be noted that dosages were higherper kilogram of body weight (see Table 1) thanthose used in the other studies reviewed here. Eachweek, teachers and parents assessed behaviorchange in 16 categories (e.g., activity, communica-tion, imitation, eye-contact, initiation, stereotypy)using the visual analogue scale (VAS), a rating scalesimilar to a Likert scale. In addition, the ABC wascompleted monthly to further assess treatmenteffects.

Whereas significant effects for secretin werefound for one of the six participants using parentVAS ratings, the opposite rating by teachersresulted in positive placebo effects. Significant pla-cebo effects were also found for four other partici-pants. These findings and clinical impressionsthrough visual inspection of the graphs led research-ers to conclude that the observed effects of secretinwere not different from placebo.

A comment should be made about a potentialconfound in the three aforementioned studies usingrepeated doses of secretin. As previously noted, inorder to eliminate carry-over effects of the drug, awashout period is recommended to allow effects to

10See Mesibov, Schopler, Schaffer, and Michal (1989) for a

discussion of the use of the CARS with adolescents and adults.

551Secretin as a Treatment

dissipate. However, this period varied from6 weeks between injections and 3 weeks betweentesting periods (Roberts et al., 2001) to 4 weeksbetween injections and weekly testing (Sponheimet al., 2002) to hours between drug administrationand weekly testing (Robinson, 2001). If carry-overeffects are present, and depending upon treatmentorder (secretin vs. placebo), results may be posi-tively or negatively skewed. Although carry-overeffects may not alter overall findings in any ofthese studies, to the extent that there are carry-over effects and raters are influenced by repeatedassessments (e.g., practice, results), this variableshould be taken into account when interpreting theresults of these studies.

RESEARCH TO IDENTIFY A SUBGROUP

OF RESPONDERS

In contrast to previous negative outcomes, arecent investigation identified a subgroup of chil-dren with autism who may benefit from secretininfusion. Kern, Miller, Evans, and Trivedi (2002)evaluated the effects of a single dose of porcinesecretin in a randomized, double-blind, placebo-con-trolled, crossover study of 19 children with autism,nine who had GI problems. Of those in the GIgroup, five had chronic diarrhea, two had a historyof chronic diarrhea in remission at the time of thestudy, and two had chronic constipation.

After the initial assessments were completed,either secretin or placebo was infused. At week 3,the alternative substance was infused, and all partic-ipants returned for follow-up at week 6. Using theABC, raters scored participants’ inappropriate andmaladaptive behaviors at baseline, week 3, andagain at week 6. Spoken language was measuredusing the MCDI. Parents provided information onGI status and global response to treatment.

Although the group with chronic diarrhea wasnot initially a target of special interest, during thestudy these children appeared to selectively respondto the secretin infusion. Therefore, their data werecompared with the data from children with no his-tory of GI problems. Although no effort had beenmade to balance the assignment of participantsacross the GI dimension initially, they were repre-sented in near-equal numbers in the secretin–pla-cebo group and the placebo–secretin group.

Results revealed significant decreases in someABC scores (Subscale I: irritability, agitation, cry-

ing; II: social withdrawal; III: stereotypy; IV: hyper-activity, non-compliance; V: inappropriate speech)in participants with chronic diarrhea when theyreceived secretin but not when they received a pla-cebo. Similarly, this group showed greater gains onthe MCDI when given secretin than when treatedwith placebo. Children without GI problems wereeither less likely than the group with chronic diar-rhea to show changes or showed no changes on thesame measures when treated with secretin.

Many studies (e.g., Carey et al., 2002; Chezet al., 2000; Coniglio et al., 2001; Lightdale et al.,2001; Molloy et al., 2002; Roberts et al., 2001;Unis et al., 2002) have cited GI problems in theirparticipants and several (Lightdale et al.; Robertset al.; Unis et al.) attempted to identify a possiblesubtype for whom secretin might prove effective.However, none found significant differences thatcould provide clinical indicators. The Kern et al.(2002) findings are noteworthy because they sug-gest there may be a subgroup of children with aut-ism whose symptoms could be diminished bysecretin. Nevertheless, they are in contrast to thestudy by Carey et al. who used the same outcomemeasure (the ABC) and reported worse scores ontwo of the same subtests (hyperactivity and inap-propriate speech) that were reported by Kern et al.to have improved.

A pilot study by Lightdale et al. (2001)attempted to replicate the Horvath et al. (1998)findings and address the issue of a possible sub-group of responders with specific GI complaints.Porcine secretion was infused in an uncontrolled,open-label trial with participants who were followedfor 5 weeks. Twenty children with autism who hadGI symptoms were videotaped at play before por-cine secretin infusion and afterward at weeks 1, 2,3, and 5. Research assistants conducted blindreviews of the videotapes using the Autism Observa-tion Scale. Language data were obtained with thePLS-3 at baseline and at the follow-up intervals. Aparent questionnaire was given at week 3 to provideinformation about perceived changes followingsecretin infusion.

Lightdale et al. (2001) found no significantincreases in language, social behaviors, or develop-mentally appropriate play skills and there were nosignificant decreases in behaviors associated withautism. In contrast to these findings, 70% of par-ents reported moderate to high change and 85% ofparents reported that they felt their children wouldbenefit from a second infusion of secretin. This

552 Esch and Carr

apparent expectancy effect again strongly under-scores the importance of conducting controlledstudies to separate the true effects of secretin fromits perceived benefits.

DISCUSSION

This article reviewed the results of 17 quantita-tive studies that investigated the efficacy of secretinas a treatment for the symptoms of autism spectrumdisorder. These symptoms include impairment insocial interaction, communication, and repetitive orstereotypic behavior.

Across these studies, approximately 600 chil-dren, ages 2–15, and 12 adults, all with diagnoseswithin the autism spectrum disorder, received someform of secretin (porcine, synthetic human, syn-thetic porcine, or homeopathic) and have beenassessed for its effects. The studies were conductedat various sites throughout the United States and inCanada, Great Britain, and Norway.

All but one of the studies (excluding the initialcase series report) failed to find a causal relationbetween secretin and the amelioration of autisticsymptoms across a variety of treatment variables(e.g., type of secretin, dosage potency, frequency),observation times (e.g., immediately after infusion,every week, or after several weeks), and participantcharacteristics (e.g., GI status, severity of autism/PDD, age, and history of adjunctive medicationuse). The effect of these variables was assessed on avariety of outcome measures including 17 differentrating scales and two standardized assessments (lan-guage, visual perception). Trained and untrainedobservers alike assessed outcomes.

In an effort to identify how secretin might haveproduced anecdotally reported improvements inchildren with autism, Connors and Crowell (1999)drew attention to the multi-ingredient compositionof the secretin preparation itself. The typical prepa-ration of secretin contains cysteine hydrochloride,added by the manufacturer to stabilize the secretin.This substance is not included in the placebo salinepreparation.

According to Connors and Crowell (1999), cys-teine hydrochloride has known neurotransmitterproperties and is active in the central and peripheralnervous systems. The authors assert that cysteinecould be responsible for the behavioral effectsattributed to secretin, a prospect of particular inter-est since the neurotransmitter properties of secretin

are unproven (Yung et al., 2001). While this possi-bility provides interesting speculation in light ofanecdotal reports of secretin’s positive and oftendramatic effects, it is less provocative given the neg-ative results obtained in 12 of the 13 placebo-con-trolled studies in this review. If the cysteine insecretin preparations were responsible for improve-ments, we would expect to see a significant differ-ence in performance between those receivingsecretin and those given a placebo (in which cyste-ine is absent from the preparation).

The brain–gut connection to explain thereported effects of secretin has been addressed inseveral of the studies reviewed here. The Kern et al.(2002) findings certainly hint at GI status as apromising lead in examining the neurophysiologyrelated to behaviors associated with autism, but thefindings generate many questions.

One issue is whether children with autism expe-rience GI problems in greater numbers than theirnon-autistic peers and, relatedly, whether autismdevelops subsequent to, or concurrent with, any GIdysfunction. In a recent review by Horvath and Per-man (2002), a striking difference was reported in theprevalence of GI symptoms between children withautism and their healthy siblings. A comparison sur-vey of 43 sibling pairs reported that 84% of thosewith autism had GI symptoms, whereas these symp-toms were present in only 31% of their non-autisticsiblings. In contrast, a recent population-basedstudy in the United Kingdom (Black, Kaye, & Jick,2002) reported equal prevalence (9%) of GI disor-ders among 96 children diagnosed with autism andtheir peers without autism (449 matched controls).

If secretin does contribute to improvement inthe symptoms of autism, its mechanism of action isunclear. It is not known whether the peptide crossesthe blood–brain barrier11 to impact the central ner-vous system directly, if at all, or contributes in someway more peripherally (e.g., improving digestion).Thus, it would be important to determine whetherGI dysfunction involves processes that include mul-tiple organs (i.e., brain and GI-related structures) ormanifests independent of, or secondarily to, the cen-tral nervous system.

At present, there is no consensus on the associ-ation between the development of autism and GI ill-ness (Black et al., 2002). In fact, several studies

11Banks, Goulet, Rusche, Niehoff, and Boismenu (2002) suggest

that it does in mice; however, in humans the properties of pos-

sible neurotransmission of secretin are still undefined.

553Secretin as a Treatment

provide evidence against a substantial association(Black et al.; Fombonne & Chakrabarti, 2001; Tay-lor et al., 2002). In contrast, Horvath and Perman(2002) cite numerous studies in the areas ofendoscopy, histology, and immunology that are allsuggestive of a GI connection to autism. Neverthe-less, they caution that as tempting as it might be toembrace this biologic link to autism, further rigor-ous investigation is crucial in providing empiricalsupport for this claim. Ultimately, it is possible thatsecretin would prove ineffective even with a verifi-able GI connection to autism.

How do we explain the continued interest ofparents in seeking secretin treatment for their chil-dren despite the inability of researchers to establishits efficacy? First, we must recognize that parents asa group are no less able to objectively evaluate theavailable scientific information than other layper-sons. However, unusual circumstances precededthese investigations. The atmosphere in which manyparents sought to participate in secretin studies washighly charged with promotion and expectations forits success. This was a unique situation since theimpetus to begin this line of research was largelystimulated by intense media attention. None of theprevious studies of other pharmacological treat-ments for autism operated in this atmosphere ofmedia attention and rapid dissemination of anec-dotal reports of success (Sandler & Bodfish, 2000).

Thus, several factors may have facilitated theinitial enthusiasm for secretin. Foremost is the moti-vational impetus by parents to find any promisingtreatment for this disorder whose symptoms areoften extremely stressful for parents. That thesesymptoms are pervasive, in contrast to the morelimited impact of other physical or mental disabili-ties (Herbert et al., 2002), may only intensify par-ents’ desperation to find relief. Dunn-Geier et al.(2000) suggest that early positive reports of the ben-efits of secretin by a parent may have increased itsinitial attractiveness and perceived utility. Hope forthe drug’s efficacy may also have been high since itis a natural substance found in the human bodyand its adverse effects appear minimal; thus, it maybe seen as relatively safe, and by extension, possiblyhelpful. In addition, there is intuitive face validityto secretin when presented in the context of thebrain–gut hypothesis.

These possibilities might result in sometimessubtle expectations that can camouflage effects indrug studies and the scientific community needs tobe mindful of their influence, particularly since

outcome measures are frequently reported by par-ents. It is incumbent upon researchers and practitio-ners to help parents and caregivers understandplacebo and expectancy effects in an effort to helpthem prevent expending resources on unproventreatments, particularly given the potential forhealth risks when applications are off-label (i.e., notFDA-approved) treatments.

Expectancy effects, however, may not whollyexplain the numerous anecdotal reports from par-ents of diminished GI symptoms, improved sleeppatterns, and better functional use of language. It isdifficult to identify the factors in the Horvath et al.(1998) case series that resulted in reports ofimproved functioning following the secretin chal-lenge for GI symptoms. Variables such as age,unique patient characteristics, anesthesia or seda-tives, duration of infusion, or onset of symptomsmay have operated singly or in some idiosyncraticcombination to have an effect when combined withsecretin administration. However, given the experi-mental evidence it seems likely that there was nospecific effect of secretin in Horvath et al. but thatother variables (possibly including informant andobserver expectancies) were responsible for thechanges reported.

Autism is a complex condition with a varietyof symptoms that casually appear to wax and wane.It is not unreasonable that such variability mightcoincide with a treatment regimen, resulting in theperception of improvement. Furthermore, childrenwith autism are not a homogeneous group. Notonly do their behaviors vary, but their responses toa variety of treatments are often diverse.

Given this, it seems likely that many differentetiologies may contribute to the behaviors seen inautism. Roberts et al. (2001) note there may beunderlying GI pathology or other factors such asdiet or anxiety responsible for symptoms of autism.Carey et al. (2002) identify GI permeability as apossible contributor to the pathogenesis of autism.Horvath and Perman (2002) review a variety oflikely suspects including abnormal metabolism inthe liver, digestive enzyme deficiency, and inflamma-tion of the upper and lower intestinal tract. Furtherscientific evidence may help solve the mystery ofhow, or if, autism relates to a brain–gut connection.At this time, however, the role that secretin plays insolving this mystery seems negligible.

The explosion of interest in secretin resulted inhundreds, and possibly thousands, of children beingexposed to this hormone. It prompted a line of

554 Esch and Carr

research whose costs could easily be measured inmillions of dollars, and whose empirical evidenceoverwhelmingly suggests there is no symptomaticrelief of autism from the hormone secretin beyondthat which could be expected from placebo.

Despite these definitive results, parents continueto be exposed to influences that offer hope for secre-tin (see Herbert et al., 2002). This makes the poten-tial cost to parents very dear indeed. Time andemotion have been, and may continue to be,expended to pursue this treatment that neverenjoyed empirically proven benefits for autism, nordoes it yet.

Three years ago, Lightdale et al. (2001)observed that the interest in secretin was likely tofade if empirical studies continued to contradict itsefficacy. Since that time, five more controlled studieshave failed to validate secretin as an effective treat-ment for autism. We submit that the findings sum-marized here should be considered conclusive andpropose that we turn our attention to guiding par-ents and caregivers toward proven treatments (e.g.,applied behavior analysis; Smith, 1999) that have ahistory of success founded on rigorous science (seeTanguay, 2002).

In the meantime, other treatments will beadvanced and claims will be made for their effi-cacy. Some of them may merit our attention; allof them may require our response. In a worldwhere technology can instantly catapult any com-munication to prominence, and pseudoscience isoften difficult to recognize, we must act responsiblyto help parents, caregivers, and others evaluate thestrengths and impact of yet unproven treatments.Above all, we must strive to advocate for rigorousscientific investigations that will define the thera-peutic benefit of such treatments before anecdotalsurrogates for those investigations wreak emotionaland financial havoc on the lives of those we endea-vor to help.

REFERENCES

Alexander, D. (1999, December). Study shows secretin fails tobenefit children with autism [Comments]. NIH News Alert,National Institutes of Health.

Aman, M. G., & Armstrong, S. A. (2000). Regarding secretin fortreating autistic disorder [Letter to the editor]. Journal ofAutism and Developmental Disorders, 30, 71–72.

American Academy of Child & Adolescent Psychiatry. (1999,March 3). Secretin in the treatment of autism. [Policy state-ment]. Retrieved June 26, 2002, from http://www.aacap.org/publications/policy/ps30.htm#TOP.

American Psychiatric Association. (1994). Diagnostic and statisti-cal manual of mental disorders (4th ed.). Washington, DC:Author.

Banks, W. A., Goulet, M., Rusche, J. R., Niehoff, M. L., &Boismenu, R. (2002). Differential transport of a secretin ana-log across the blood–brain and blood–cerebrospinal fluidbarriers of the mouse. The Journal of Pharmacology andExperimental Therapeutics, 302, 1062–1069.

Bertrand, J., Mars, A., Boyle, C., Bove, F., Yeargin-Allsopp, M.,& Decoufle, P. (2001). Prevalence of autism in a UnitedStates population: The Brick Township, New Jersey, investi-gation. Pediatrics, 108, 1155–1161.

Black, C., Kaye, J. A., & Jick, H. (2002). Relation of childhoodgastrointestinal disorders to autism: Nested case-controlstudy using data from the UK General Practice ResearchDatabase. British Medical Journal, 325, 419–421.

Carey, T., Ratliff-Schaub, K., Funk, J., Weinle, C., Myers, M., &Jenks, J. (2002). Double-blind placebo-controlled trial ofsecretin: Effects on aberrant behavior in children with aut-ism. Journal of Autism and Developmental Disorders, 32,161–167.

Chez, M. G., & Buchanan, C. P. (2000). Reply to B. Rimland’s‘‘comments on ‘secretin and autism: A two-part clinicalinvestigation.’ ’’ Journal of Autism and Developmental Disor-ders, 30, 97.

Chez, M. G., Buchanan, C. P., Bagan, B. T., Hammer, M. S.,McCarthy, K. S., Ovrutskaya, I., et al. (2000). Secretin andautism: A two-part clinical investigation. Journal of Autismand Developmental Disorders, 30, 87–94.

Christ, A., Werth, B., Hildebrand, P., Gyr, K., Stalder, G. A., &Beglinger, C. (1988). Human secretin: Biologic effects andplasma kinetics in humans. Gastroenterology, 94, 311–316.

Coniglio, S., Lewis, J. D., Lang, C., Burns, T. G., Subhani-Siddique, R., Weintraub, A., et al. (2001). A randomized,double-blind, placebo-controlled trial of single-doseintravenous secretin as treatment for children with autism.The Journal of Pediatrics, 138, 649–655.

Connors, S. L., & Crowell, D. E. (1999). Secretin and autism:The role of cysteine [Letter to the editor]. Journal of theAmerican Academy of Child and Adolescent Psychiatry, 38,795–796.

Corbett, B., Khan, K., Czapansky-Beilman, D., Brady, N.,Dropik, P., Zelinsky Goldman, D., et al. (2001). A double-blind, placebo-controlled crossover study investigating theeffect of porcine secretin in children with autism. ClinicalPediatrics, 40, 327–331.

DiCicco-Bloom, E. (1998). Invited commentary. Newsletter of theNational Alliance for Autism Research, 4, 24.

Dunn-Geier, J., Ho, H. H., Auersperg, E., Doyle, D., Eaves, L.,Matsuba, C., et al. (2000). Effect of secretin on children withautism: A randomized controlled trial. Developmental Medi-cine and Child Neurology, 42, 796–802.

Fombonne, E., & Chakrabarti, S. (2001). No evidence for a newvariant of measles-mumps-rubella-induced autism. Pediatrics,108, Article e58. Retrieved August 18, 2003, from http://www.pediatrics.org/cgi/content/full/108/4/e58.

Hardman, M. L., Drew, C. J., & Egan, M. W. (1999). Humanexceptionality: Society, school, and family (6th ed.). Boston:Allyn and Bacon.

Herbert, J. D., Sharp, I. R., & Gaudiano, B. A. (2002). Separat-ing fact from fiction in the etiology and treatment of autism:A scientific review of the evidence. The Scientific Review ofMental Health Practice, 1, 23–43. Retrieved December 9,2002, from http://www.scientificmentalhealth.org/SRMHP/current.html.

Hirsch, D. (1999). Secretin in the treatment of autism. TheExceptional Parent, 29, 94.

Horvath, K. (2000). Secretin treatment for autism [Letter to theeditor]. The New England Journal of Medicine, 342, 1216.

555Secretin as a Treatment

Horvath, K., Papadimitriou, J. C., Rabsztyn, A.,Drachenberg, C.,& Tildon, J. T. (1999). Gastrointestinal abnormalities in chil-dren with autistic disorder. Journal of Pediatrics, 135,559–563.

Horvath, K., & Perman, J. A. (2002). Autism and gastrointestinalsymptoms. Current Gastroenterology Reports, 4, 251–258.

Horvath, K., Stefanatos, G., Sokolski, K. N., Watchel, R.,Nabors, L., & Tildon, J. T. (1998). Improved social and lan-guage skills after secretin administration in patients withautistic spectrum disorders. Journal of the Association forAcademic Minority Physicians, 9, 9–15.

Kern, J. K., Miller, V. S., Evans, P. A., & Trivedi, M. H. (2002).Efficacy of porcine secretin in children with autism and per-vasive developmental disorder. Journal of Autism and Devel-opmental Disorders, 32, 153–160.

Lightdale, J. R., Hayer, C., Duer, A., Lind-White, C., Jenkins, S.,Siegel, B., et al. (2001, November). Effects of intravenoussecretin on language and behavior of children with autism andgastrointestinal symptoms: A single-blinded, open-label pilotstudy. Pediatrics, 108, Article e90. Retrieved June 11, 2002,from http://www.pediatrics.org/cgi/content/full/108/5/e90.

McMillin, D. L., Richards, D. G., Mein, E. A., & Nelson, C. D.(1999). The abdominal brain and enteric nervous system.The Journal of Alternative and Complementary Medicine, 5,575–586.

McQueen, J. M., & Heck, A. M. (2002). Secretin for the treatmentof autism. The Annals of Pharmacotherapy, 36, 305–311.

Mesibov, G., Schopler, E., Schaffer, B., & Michal, N. (1989). Useof the Childhood Autism Rating Scale with autistic adoles-cents and adults. Journal of the American Academy of Childand Adolescent Psychiatry, 28, 538–541.

Molloy, C. A., Manning-Courtney, P., Swayne, S., Bean, J.,Brown, J. M., Murray, D. S., et al. (2002). Lack of benefitof intravenous synthetic human secretin in the treatment ofautism. Journal of Autism and Developmental Disorders, 32,545–551.

National Institute of Child Health and Human Development.(1999, December 8). [National Institutes of Health NewsAlert]. Study shows secretin fails to benefit children with aut-ism. Retrieved June 10, 2002, from http://www.nichd.nih.-gov/new/releases/secretin1299.htm.

Owley, T., McMahon, W., & Cook, E. H. (2001). Multisite, dou-ble-blind, placebo-controlled trial of porcine secretin in aut-ism. Journal of the American Academy of Child andAdolescent Psychiatry, 40, 1293–1299.

Owley, T., Steele, E., Corsello, C., Risi, S., McKaig, K.,Lord, C., et al. (1999, October 6). A double-blind, placebo-controlled trial of secretin for the treatment of autistic disor-der. Medscape General Medicine, 1, 1–8. Retrieved June 12,2002, from http://medscape.com/viewarticle/408013.

Rimland, B. (1999, January). The use of secretin in autism: somepreliminary answers. Autism Research Review International.Retrieved June 1, 2002, from http://www.autism.com/ari/newslet.html.

Rimland, B. (2000). Comments on ‘‘Secretin and autism: A two-part clinical investigation’’ by M. G. Chez et al. Journal ofAutism and Developmental Disorders, 30, 95.

Roberts, W., Weaver, L., Brian, J., Bryson, S., Emelianova, S.,Griffiths, A.-M., et al. (2001, May). Repeated doses of por-cine secretin in the treatment of autism: A randomized, pla-cebo-controlled trial. Pediatrics, 107, Article e71. RetrievedJune 11, 2002, from http://www.pediatrics.org/cgi/content/full/107/5/e71.

Robinson, T. W. (2001). Homeopathic secretin in autism: A clini-cal pilot study. British Homeopathic Journal, 90, 86–91.

Sandler, A. D., & Bodfish, J. W. (2000). Placebo effects in aut-ism: Lessons from secretin. Developmental and BehavioralPediatrics, 21, 347–350.

Sandler, A. D., Sutton, K. A., DeWeese, J., Girardi, M. A.,Sheppard, V., & Bodfish, J. W. (1999). Lack of benefit of asingle dose of synthetic human secretin in the treatment ofautism and pervasive developmental disorder. New EnglandJournal of Medicine, 341, 1801–1806.

Shutt, C. E. (1998). Secretin and autism: A clue but not a cure.Newsletter of the National Alliance for Autism Research, 4,21–24.

Smith, T. (1999). Outcome of early intervention for childrenwith autism. Clinical Psychology: Science and Practice, 6,33–49.

Sponheim, E., Oftedal, G., & Helverschou, S. B. (2002). Multipledoses of secretin in the treatment of autism: A controlledstudy. Acta Pædiatrica, 91, 540–545.

Tanguay, P. E. (2002). Commentary: The primacy of the scien-tific method. Journal of the American Academy of Child andAdolescent Psychiatry, 41, 1322–1323.

Taylor, B., Miller, E., Lingam, R., Andrews, N., Simmons, A., &Stowe, J. (2002). Measles, mumps, and rubella vaccinationand bowel problems or developmental regression in childrenwith autism: population-based study. British Medical Jour-nal, 324, 393–396.

Unis, A. S., Munson, J. A., Rogers, S. J., Goldson, E.,Osterling, J., Gabriels, R., et al. (2002). A randomized, dou-ble-blind, placebo-controlled trial of porcine versus syntheticsecretin for reducing symptoms of autism. Journal of theAmerican Academy of Child and Adolescent Psychiatry, 41,1315–1321.

Veenstra-Vanderweele, J., & Cook, E. H., Jr. (2003). Genetics ofchildhood disorders: XLVI. Autism, part 5: Genetics of aut-ism. Journal of the American Academy of Child and Adoles-cent Psychiatry, 42, 116–118.

Volkmar, F. (1999). Lessons from secretin [Commentary]. TheNew England Journal of Medicine, 341, 1842–1844.

Volkmar, F. (2000). [Letter to the editor]. The New England Jour-nal of Medicine, 342, 1218.

Yung, W.-H., Leung, P.-S., Ng, S. S. M., Zhang, J., Chan, S. C.Y., & Chow, B. K. C. (2001). Secretin facilitates GABA trans-mission in the cerebellum. The Journal of Neuroscience, 21,7063–7068.

556 Esch and Carr