Nonpharmacologic Interventions for Treatment-Resistant Depression in Adults Executive Summary

Transcription

1 Comprtive Effectiveness Review Numer 33 Effective Helth Cre Progrm Nonphrmcologic Interventions for Tretment-Resistnt Depression in Adults Executive Summry Bckground Mjor depressive disorder (MDD) is common nd costly. Over the course of yer, etween 13.1 million nd 14.2 million people will experience MDD. Approximtely hlf of these people seek help for this condition, nd only 20 percent of those receive dequte tretment. For those who do initite tretment for their depression, pproximtely 50 percent will not dequtely respond following cutephse tretment; this refrctory group hs considerle clinicl nd reserch interest. Ptients with only one prior tretment filure re sometimes included in this group, ut ptients with two or more prior tretment filures re prticulrly importnt nd poorly understood group nd re considered to hve tretment-resistnt depression (TRD). These TRD ptients represent complex popultion with disese tht is difficult to mnge. Ptients with TRD incur the highest direct nd indirect medicl costs mong those with MDD. These costs increse with the severity of TRD. Tretment-resistnt ptients re twice s likely to e hospitlized, nd their cost of hospitliztion is more thn six times the men totl costs of depressed ptients who re not tretment resistnt. After Effective Helth Cre Progrm The Effective Helth Cre Progrm ws initited in 2005 to provide vlid evidence out the comprtive effectiveness of different medicl interventions. The oject is to help consumers, helth cre providers, nd others in mking informed choices mong tretment lterntives. Through its Comprtive Effectiveness Reviews, the progrm supports systemtic pprisls of existing scientific evidence regrding tretments for high-priority helth conditions. It lso promotes nd genertes new scientific evidence y identifying gps in existing scientific evidence nd supporting new reserch. The progrm puts specil emphsis on trnslting findings into vriety of useful formts for different stkeholders, including consumers. The full report nd this summry re ville t hrq.gov/reports/finl.cfm. considering oth medicl nd disility clims from n employer s perspective, one study found tht TRD employees cost $14,490 per employee per yer, wheres Effective Helth Cre

2 the cost for non-trd employees ws $6,665 per employee per yer. Given the urden of TRD generlly, the uncertin prognosis of the disorder, nd the high costs of therpy, clinicins nd ptients like need cler evidence to guide their tretment decisions. The choices re wide rnging, include oth phrmcologic nd nonphrmcologic interventions, nd re frught with incomplete, potentilly conflicting evidence. Somtic tretments, which my involve use of phrmcologic intervention or device, re commonly considered for ptients with TRD. Antidepressnt medictions, which re the most commonly used intervention, hve decresing efficcy for producing remission fter ptients hve experienced two tretment filures. Such drugs lso often hve side effects, sometimes minor ut sometimes quite serious. For these resons, clinicins often look for lterntive strtegies for their TRD ptients. This review from the RTI Interntionl University of North Crolin t Chpel Hill Evidence-sed Prctice Center (EPC) provides comprehensive summry of the ville dt ddressing the comprtive effectiveness of four nonphrmcologic tretments s therpies for ptients with TRD: electroconvulsive therpy (ECT), repetitive trnscrnil mgnetic stimultion (rtms), vgus nerve stimultion (VNS), nd cognitive ehviorl therpy or interpersonl psychotherpy (CBT or IPT). The core ptient popultion of interest ws ptients with MDD who met our definition of TRD: filure to respond following two or more dequte ntidepressnt tretments. We lso included TRD studies in which the ptient popultion could include mix of up to 20 percent of ptients with ipolr disorder (i.e., 80 percent or more of ptients hd only MDD), ssuming tht this smll mix would not sustntilly lter outcomes seen with MDD-only popultions. We structured our review to mintin our focus on study popultions meeting our TRD definition ( 2 ntidepressnt filures) while not excluding potentilly relevnt evidence. We identified different tiers of TRDrelted studies to use in our nlytic strtegy: Tier 1 evidence (TRD s defined in this report): studies in which ptients specificlly hd two or more prior tretment filures with medictions. Tier 2 evidence: studies in which ptients hd one or more prior tretment filures. Tier 3 evidence: studies in which the numer of prior filed tretments ws not specified ut the clinicl sitution suggested high proility of ptients hving two or more prior ntidepressnt tretment filures; these dt hve prole relevnce to TRD. Studies tht did not specify the numer of filed tretments ut noted tht ll sujects were referred for ECT were included in this tier. This comprtive effectiveness review is intended to help vrious decisionmkers come to informed choices out the use of nonphrmcologic interventions for TRD in dults. Our principl gol is to summrize comprtive dt on the efficcy, effectiveness, nd hrms of ECT, rtms, VNS, nd CBT/IPT in ptients with TRD. Comprisons of these nonphrmcologic therpies re our min interest. However, ecuse tretment decisions mde y ptients with TRD nd their clinicins re not limited to nonphrmcologic options, we lso compre nonphrmcologic options with phrmcologic ones. We ddress the following six Key Questions (KQs) s specified y the Agency for Helthcre Reserch nd Qulity (AHRQ). Trils in these KQs refers to tretment ttempts, not experimentl studies. KQ 1. For dults with TRD (defined s two or more filed dequte trils of iologic [i.e., phrmcologic] intervention), do nonphrmcologic interventions such s electroconvulsive therpy (ECT), repetitive trnscrnil mgnetic stimultion (rtms), vgus nerve stimultion (VNS), or demonstrted effective psychotherpy (e.g., cognitive therpy [CBT or IPT]) differ in efficcy or effectiveness in treting cute-phse depressive symptoms (e.g., response nd remission), whether s single tretment or prt of comintion tretment? KQ 1. How do these nonphrmcologic tretments compre with phrmcologicl tretments in efficcy or effectiveness in treting cute-phse depressive symptoms fter two or more filed dequte trils? 2

3 KQ 2. For dults with TRD, do nonphrmcologic interventions differ in their efficcy or effectiveness for mintining response or remission (e.g., preventing relpse or recurrence), whether s single tretment or prt of comintion tretment? KQ 3. Do nonphrmcologic interventions (single or comintion) differ in their efficcy or effectiveness for treting TRD s function of prticulr symptom sutypes (e.g., cttonic [frozen or hyper] or psychotic symptoms)? KQ 4. For dults with TRD, do nonphrmcologic interventions differ in sfety, dverse events, or dherence? Adverse effects of interest include ut re not limited to mnesi, memory loss, hedches, nd postopertive complictions. KQ 5. How do the efficcy, effectiveness, or hrms of tretment with nonphrmcologic tretments for TRD differ for the following supopultions: Elderly or very elderly ptients; other demogrphic groups (defined y ge, ethnic or rcil groups, nd sex)? Ptients with medicl comoridities (e.g., seizure history, stroke, dietes, dementi, perintl depression, ischemic hert disese, cncer)? KQ 6. For dults with TRD, do nonphrmcologic interventions differ in regrd to other helthrelted outcomes (e.g., qulity of life)? We serched MEDLINE, Emse, the Cochrne Lirry, PsycINFO, nd Interntionl Phrmceuticl Astrcts. We serched for systemtic reviews, clinicl controlled trils, met-nlyses, nd nonexperimentl studies in which the investigtor did not ssign group lloction. Sources were serched from 1980 through Novemer 18, AHRQ Scientific Resource Center (SRC) stff contcted device mnufcturers nd invited them to sumit dossiers, including cittions. The SRC lso provided our EPC with other relevnt dt tht my not hve een cptured in the literture serch. For efficcy nd effectiveness (KQs 1 nd 2), we first focused on hed-to-hed rndomized controlled trils (RCTs) compring one intervention with nother. When sufficient hed-to-hed evidence ws unville, we evluted indirect evidence: nonphrmcologic interventions versus plceo- or shm-controlled evidence or tretment s usul controls. For KQs 3, 4, 5, nd 6, we exmined dt from oth experimentl nd oservtionl studies (generlly prospective cohort studies). We did not formlly distinguish efficcy from effectiveness trils. We rted the qulity of individul studies s good, fir, or poor; only good or fir studies re included in these nlyses. We evluted the strength of the vrious odies of evidence using principles stted in the AHRQ Methods Guide for Comprtive Effectiveness Reviews, which grdes strength s high, moderte, low, or insufficient. We evluted the pplicility of the ody of evidence using qulittive ssessment of the popultion, intervention/tretment, comprtor, outcomes mesured, timing of followup, nd setting. Throughout this report we synthesized the literture qulittively. If dt were sufficient, we conducted met-nlyses of dt for comprisons involving trils tht were firly homogenous in study popultions, tretment intervention, nd outcome ssessments. Given our focus on Tier 1 (TRD) studies, for ech KQ we first present n overview of the prticulr comprison, including the strength of evidence findings for the Tier 1 studies. This summry does not present detiled findings from the Tier 2 nd Tier 3 studies. The results chpter of the full report presents those dt in greter detil. Results: Overview From totl of 2,754 cittions retrieved, we ultimtely identified 79 good-, fir-, or poor-qulity rticles in this review; they represent 64 studies. Of these studies, there were 17 hed-to-hed RCTs (19 rticles): 7 studies (9 rticles) were hed-to-hed RCTs of nonphrmcologic intervention versus nonphrmcologic intervention; 3 were hed-to-hed RCTS of nonphrmcologic intervention versus phrmcologic one; nd 7 were hed-to-hed studies of 3

4 phrmcologic versus phrmcologic intervention. Further, there were 38 dditionl RCTs (50 rticles) tht were shm- or plceo-controlled, nd 2 oservtionl studies (2 rticles). We excluded 8 studies (8 rticles) ecuse of poor qulity. We present evidence tht llows comprison of the four nonphrmcologic tretments of interest (ECT, rtms, VNS, nd psychotherpy) strtified y tiers of evidence. Comprtive clinicl reserch on nonphrmcologic interventions in TRD popultion is in its infncy. Mny clinicl questions out efficcy nd effectiveness remin unnswered. The text elow presents our principl results; summry tles (A J) document Tier 1 TRD findings for mjor comprisons nd outcomes for ech key question, give the overll strength of evidence for tht comprison, nd outline key findings. We report first on direct evidence (hedto-hed comprisons) nd then on indirect evidence (e.g., trils using controls). If specific comprison did not involve Tier 1 popultion ut did hve trils conducted in Tier 2 nd/or Tier 3 popultion, we hve listed it in this tle, noted No eligile studies identified, nd dded footnote indicting the presence of t lest one such study. The gretest volume of evidence is for ECT nd rtms; however, the direct comprtive evidence out even these tretments is quite limited. Aville indirect evidence primrily involves rtms; little informtion is ville on VNS nd psychotherpy (chiefly for efficcy nd dverse events), nd no ville indirect evidence involves ECT. Given the limited numer of Tier 1 studies incomplete reporting on the numer of filed tretment ttempts, we were unle to strtify our outcomes y the numer of tretment filures within Tier 1. 4

5 Tle A. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for Key Question 1, comprtive efficcy of nonphrmcologic tretments ECT vs. rtms Chnge in 42 Low 1 fir tril: oth ECT nd rtms improved depressive severity symptom ut did not differ significntly. ECT vs. rtms Response rte 42 Low 1 fir tril: ECT nd rtms did not differ significntly. ECT vs. rtms Remission rte 42 Low 1 fir tril: ECT nd rtms did not differ significntly. ECT plus rtms Chnge in 22 Low 1 fir tril: oth ECT nd ECT plus rtms vs. ECT depressive severity improved symptom severity ut did not differ significntly. ECT plus rtms Response rte 0 NA No eligile studies identified. c vs. ECT ECT plus rtms Remission rte 22 Low 1 fir tril: ECT nd ECT plus rtms did not vs. ECT differ significntly. ECT vs. shm Chnge in 0 NA No eligile studies identified. c depressive severity ECT vs. shm Response rte 0 NA No eligile studies identified. c ECT vs. shm Remission rte 0 NA No eligile studies identified. c rtms vs. shm Chnge in 497 High 7 trils (3 good, 4 fir): rtms hd depressive severity significntly greter decrese in depressive severity thn shm. 4 fir trils: rtms hd nonsignificntly greter decrese in depressive severity thn shm. 2 fir trils: rtms hd greter decrese thn shm ut significnce NR. 1 fir tril: rtms did not significntly differ from shm. rtms vs. shm Response rte 471 High 4 trils (3 good, 1 fir): rtms hd significntly higher response rte thn shm. 1 fir tril: rtms hd nonsignificntly higher response rte thn shm. 6 fir trils: rtms hd higher response rte thn shm, ut significnce NR. 1 fir tril: rtms did not clerly differ from shm, ut significnce NR. rtms vs. shm Remission rte 223 Moderte 3 trils (2 good, 1 fir): rtms hd significntly greter remission rte thn shm. 1 fir tril: rtms hd greter remission rte thn shm ut significnce NR. 5

6 Tle A. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for Key Question 1, comprtive efficcy of nonphrmcologic tretments (continued) VNS vs. shm Chnge in 235 Low 1 good tril: VNS nd shm did not differ severity depressive significntly. VNS vs. shm Response rte 235 Low 1 good tril: VNS nd shm did not differ significntly. Psychotherpy Chnge in 0 NA No eligile studies identified. c vs. control depressive severity Psychotherpy Response rte 0 NA No eligile studies identified. c vs. control Psychotherpy Remission rte 0 NA No eligile studies identified. c vs. control ECT = electroconvulsive therpy; NA = not pplicle; NR = not reported; rtms = repetitive trnscrnil mgnetic stimultion; VNS = vgus nerve stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 or Tier 3 study ddressed this comprison. Tle B. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 1, comprtive efficcy of nonphrmcologic nd phrmcologic tretments ECT vs. Chnge in 39 Low 1 fir tril: ECT hd significntly greter phrmcotherpy depressive severity improvement in symptom severity thn phrmcotherpy. ECT vs. Response rte 39 Low 1 fir tril: ECT hd significntly greter response rtes thn phrmcotherpy. phrmcotherpy Psychotherpy vs. Chnge in 0 NA No eligile studies identified. c phrmcotherpy depressive severity Psychotherpy vs. Response rte 0 NA No eligile studies identified. c phrmcotherpy Psychotherpy vs. Remission rte 0 NA No eligile studies identified. c phrmcotherpy ECT = electroconvulsive therpy; NA = not pplicle; rtms = repetitive trnscrnil mgnetic stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. 6

7 Tle C. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 2, comprtive efficcy for mintining remission ECT vs. rtms Mintennce of 0 NA No eligile studies identified. c remission rtms vs. shm Mintennce of 68 Insufficient 3 fir trils: no significnt differences in remission mintennce of remission; however, smll smple sizes in two of the studies nd the presence of co-intervention in the third study mke results difficult to interpret. CBT vs. usul Mintennce of 0 NA No eligile studies identified. c cre remission CBT = cognitive ehviorl therpy; ECT = electroconvulsive therpy; NA = not pplicle; rtms = repetitive trnscrnil mgnetic stimultion; vs = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. Tle D. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 3, comprtive efficcy for prticulr symptom sutypes ECT vs. rtms Chnge in 0 NA No eligile studies identified. c depressive severity ECT = electroconvulsive therpy; NA = not pplicle; rtms = repetitive trnscrnil mgnetic stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. 7

8 Tle E. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 4, impct of nonphrmcologic interventions on cognitive functioning ECT vs. rtms Cognitive 72 Insufficient 1 fir tril nd 1 fir cohort study: Some functioning evidence suggests no difference etween tretments, wheres some evidence suggests ECT my hve deleterious impct on cognitive functioning compred with rtms (1 study: significnt effect on 1-week recll; oth studies: nonsignificnt effect on ll other mesures). ECT vs. Cognitive 22 Insufficient 1 fir tril: no significnt differences in ECT + rtms functioning single item mesure on memory prolems. rtms vs. shm Cognitive 161 Insufficient 4 trils (1 good, 3 fir): Some evidence functioning suggests no difference etween rtms nd shm, wheres some evidence suggests tht rtms improves cognitive functioning compred to shm (2 trils: significnt differences in memory, verl fluency; ll other findings nonsignificnt or significnce not reported). ECT = electroconvulsive therpy; rtms = repetitive trnscrnil mgnetic stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. Tle F. Summry of findings on nonphrmcologic tretment of dult tretment-resistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 4, specific dverse events ECT vs. rtms Adverse events 0 NA No eligile studies identified. c ECT vs. Adverse events 22 Low 1 fir tril: no significnt differences in ECT + rtms specific dverse events rtms vs. shm Adverse events 68 Low 1 good tril: rtms resulted in significntly more sclp pin t the stimultion site thn shm. VNS vs. shm Adverse events 235 Low 1 fir tril: Some differences in specific dverse events reported (P = NR) ECT = electroconvulsive therpy; NA = not pplicle; rtms = repetitive trnscrnil mgnetic stimultion; VNS = vgus nerve stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. 8

9 Tle G. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 4c, withdrwls due to dverse event ECT vs. rtms Withdrwls 30 Low 1 fir cohort study: no difference in withdrwls etween ECT nd rtms groups (P = NR). ECT vs. shm Withdrwls 0 NA No eligile studies identified. c rtms vs. shm Withdrwls 337 Insufficient 7 trils (1 good, 6 fir): trils showed mixed results out withdrwls ttriuted to dverse events. VNS vs. shm Withdrwls 235 Low 1 good tril: VNS hd greter withdrwls ttriuted to dverse events thn shm (significnce NR). CBT vs. usul Withdrwls 0 NA No eligile studies identified. c cre CBT = cognitive ehviorl therpy; ECT = electroconvulsive therpy; NA = not pplicle; NR = not reported; rtms = repetitive trnscrnil mgnetic stimultion; VNS = vgus nerve stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. Tle H. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 4d, dherence s mesured y overll withdrwls ECT vs. rtms Overll withdrwls 72 Low 1 fir tril nd 1 fir cohort study: studies showed more withdrwls in the ECT group compred with rtms (P = NR). ECT vs. shm Overll withdrwls 0 NA No eligile studies identified. c rtms vs. shm Overll withdrwls 325 Insufficient 8 fir trils: trils showed mixed results out withdrwls. CBT vs. usul Overll withdrwls 0 NA No eligile studies identified. c cre CBT = cognitive ehviorl therpy; ECT = electroconvulsive therpy; NA = not pplicle; rtms = repetitive trnscrnil mgnetic stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. 9

10 Tle I. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 5, efficcy nd hrms for selected popultions rtms vs. shm Chnges in 34 Low 1 fir tril: rtms produced etter outcome depressive severity thn shm in young dult popultion (ges 18 37). rtms vs. shm Chnges in 20 Low 1 fir tril: rtms produced etter outcome depressive severity thn shm in older dults with post-stroke depression. rtms vs. shm Response 34 Low 1 fir tril: rtms produces etter response rtes thn shm in young dult popultion (ges 18 37). rtms vs. shm Response 20 Low 1 fir tril: no difference etween rtms nd shm for older dults with post-stroke depression. rtms vs. shm Remission 20 Low 1 fir tril: no difference etween rtms nd shm in older dults with post-stroke depression. rtms = repetitive trnscrnil mgnetic stimultion; vs. = versus Strength of evidence is sed on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. Tle J. Summry of findings on nonphrmcologic tretment of dult tretmentresistnt depression (TRD) with strength of evidence for Tier 1 (TRD) for KQ 6, helthrelted outcomes ECT vs. Helth-relted 22 Low 1 fir tril: There were no differences ECT + rtms outcomes etween groups in improvements in dily functioning. rtms vs. shm Helth-relted 60 Low 1 fir tril: low rtms hd significntly outcomes greter improvement in helth sttus nd dily functioning thn shm, while this reltionship pproched sttisticl significnce when compring high rtms to shm. VNS vs. shm Helth-relted 214 Low 1 fir tril: VNS nd shm groups did not outcomes differ significntly in dily functioning. CBT/DBT vs. Helth-relted 0 NA No eligile studies identified. c control outcomes CBT = cognitive ehviorl therpy; DBT = dilecticl ehviorl therpy; NA = not pplicle; rtms = repetitive trnscrnil mgnetic stimultion; VNS = vgus nerve stimultion; vs. = versus Strength of evidence is sed on the on guidnce provided in the AHRQ Methods Guide for Comprtive Effectiveness Reviews; see text. c At lest one Tier 2 nd/or Tier 3 study ddressed this comprison. 10

11 Efficcy of Nonphrmcologic Interventions Aginst Other Nonphrmcologic Interventions (KQ 1) Direct evidence. The ville hed-to-hed literture concerning the efficcy of the nonphrmcologic interventions for Tier 1 TRD is limited to two fir trils (oth in MDD-only popultions). One compred ECT nd rtms, nd the other compred ECT nd ECT plus rtms. They showed, with low strength of evidence, no differences etween tretment options for depressive severity, response rtes, nd remission rtes. No tril involved direct comprison of psychotherpy with nother nonphrmcologic intervention. Indirect evidence. We identified trils tht compred nonphrmcologic intervention, generlly rtms, VNS, or psychotherpy, with control or shm procedure in Tier 1 popultions. We identified no eligile ECT versus control studies. The numer of these trils with the sme or similr control group ws very smll, so we could not pool them quntittively. We could, however, ssess the potentil enefits of nonphrmcologic interventions versus controls y clculting men chnges in depressive severity, reltive risks of response, nd reltive risks of remission. rtms ws eneficil reltive to controls receiving shm procedure for ll three outcomes (severity of depressive symptoms, response rte, remission rte). rtms produced greter decrese in depressive severity (high strength of evidence). Specificlly, rtms verged decrese in depressive severity mesured y the Hmilton Rting Scle for Depression (HAM-D) of more thn 5 points reltive to shm control, nd this chnge meets the minimum threshold of the 3-point HAM-D difference tht is considered cliniclly meningful. Response rtes were greter with rtms thn shm (lso high strength of evidence); those receiving rtms were more thn three times s likely to chieve depressive response s ptients receiving shm procedure. Finlly, rtms ws lso more likely to produce remission thn the control procedure (moderte strength of evidence); ptients receiving rtms were more thn six times s likely to chieve remission s those receiving the shm. In the only other Tier 1 comprison, one good-qulity VNS versus shm control tril ( mixed MDD/ipolr popultion) reported no differences etween the groups s mesured y chnge in depressive severity or response rtes (low strength of evidence). Efficcy of Nonphrmcologic Interventions Compred With Antidepressnt Phrmcotherpies (KQ 1) Direct evidence. The ville hed-to-hed literture concerning the efficcy of the nonphrmcologic interventions compred with phrmcologic tretment (in this cse, proxetine) for Tier 1 trils is limited to one fir tril ( mixed MDD/ipolr popultion). ECT produced significntly greter decrese in depressive severity (9 points y HAM-D) nd significntly etter response rtes (71 percent vs. 28 percent) thn medictions (low strength of evidence). Indirect evidence. Indirect evidence out procedures or psychotherpy (vs. shm or nonphrmcologic controls) ws presented ove s prt of KQ 1. We ttempted to determine men chnges in depressive severity, reltive risks of response, nd reltive risks of remission for phrmcologic versus control studies to llow comprison with similr outcomes in the nonphrmcologic versus control trils (KQ 1, indirect). However, we found no comprle, common control groups (i.e., ptients not receiving moodrelted mediction) to llow such comprisons. Insted, we determined men verge outcomes for phrmcologic tretments. For switching strtegies, men phrmcologic response rtes verged 39.8 percent (95% CI, 30.7% to 48.9%) nd men remission rtes verged 22.3 percent (95% CI, 16.2% to 28.4%). For ugmenttion, men response rtes verged 38.1 percent (31.0% to 45.3%) nd men remission rtes verged 27.2 percent (20.4% to 34.0%). For mintennce strtegies, men response rtes verged 27.3 percent (19.8% to 34.8%) nd men remission rtes verged 16.8 percent (13.5% to 20.2%). 11

12 Although these results provide n ide of the generl degree of response seen with next-step phrmcologic tretment in TRD, they serve s n uncontrolled cse series nd should e compred to nonphrmcologic outcomes only with cution. Mintennce of Remission or Prevention of Relpse (KQ 2) Direct evidence. With respect to mintining remission (or preventing relpse), we hd no direct comprisons involving ECT, rtms, VNS, or CBT. Indirect evidence. Three fir trils compred rtms with shm procedure nd found no significnt differences. However, too few ptients were followed during the relpse prevention phses in two of the three studies, nd ptients in the third received cointervention providing insufficient evidence for conclusion. We hd no eligile studies for ECT, VNS, or psychotherpy. Efficcy of Nonphrmcologic Interventions for Ptients With Different Symptomtology (KQ 3) Direct evidence. We identified no Tier 1 trils tht ddressed whether procedure-sed tretments differed s function of symptom sutypes. Also, no comprtive evidence ws ville out psychotherpy in sugroups defined y symptom clusters. Indirect evidence. We identified no studies testing either procedure-sed or psychotherpeutic interventions ginst shm procedures or other controls. Sfety, Adverse Events, nd Adherence (KQ 4) Direct evidence. In exmining sfety, dverse events, nd dherence, we found some differences cross the interventions in the hrms nd negtive side effects to ptients. However, the dt were insufficient to rech conclusive result. For just this set of nlyses, we exmined oth clinicl trils nd cohort studies, nd we focus on cognitive functioning, occurrence of specific dverse events, nd withdrwls. Cognitive functioning. For Tier 1 studies on cognitive functioning, some evidence suggests no differences in chnges in cognitive functioning etween groups, while some evidence suggests ECT my hve deleterious impct on cognitive functioning compred to rtms (insufficient strength of evidence). No differences etween groups on single-item mesure of cognitive functioning were found in study compring ECT with ECT nd rtms (insufficient strength of evidence). Specific dverse events. One Tier 1 study compring ECT with comintion of ECT nd rtms found no differences in specific dverse events (low strength of evidence). Withdrwls. We looked t oth withdrwls tht investigtors ttriuted to dverse events nd overll numers or rtes of withdrwls. A single study with smll smple size indicted no difference in withdrwls due to dverse events for the ECT group when compred to rtms ut did not report on the significnce of this result (low strength of evidence). Evidence for ECT compred with rtms indicted higher rtes of overll withdrwls in the ECT compred to the rtms group (P = NR; low strength of evidence). Indirect evidence. We ttempted to include dt from the sme types of studies nd for the sme outcomes s for direct evidence. We identified no studies compring ECT versus control. Cognitive functioning. Mixed evidence on cognitive functioning in rtms versus shm ws insufficient evidence to drw conclusion (insufficient strength of evidence). Specific dverse events. rtms groups reported significntly more sclp pin t the stimultion site (low strength of evidence). Some differences in the frequency of specific dverse events were seen when compring VNS nd shm groups, ut the significnce of the findings ws not reported (P = NR) (low strength of evidence). Withdrwls. Findings were mixed in Tier 1 studies s to whether rtms groups hd greter rtes of withdrwls (overll nd due to dverse events) thn groups receiving shm procedures (insufficient evidence for oth). 12

13 Withdrwls ttriutle to dverse events were higher in the VNS group compred with shm (low strength of evidence). No Tier 1 studies reported on withdrwls for CBT groups versus those receiving some form of usul cre. Efficcy or Hrms of Nonphrmcologic Tretments for Selected Ptient Sugroups (KQ 5) Direct evidence. We found no studies (in ny tier) directly compring nonphrmcologic interventions in selected popultions, such s the elderly, those with stroke, or those with other medicl comoridities. Indirect evidence. Two Tier 1 trils compred rtms with shm. All findings provided low strength of evidence. For young dults (ges 18 37), one tril found tht rtms produced greter decrese in depressive severity nd greter response rte thn shm. A second tril, conducted in older dults with post-stroke depression, found tht rtms produced greter decrese in depressive severity nd greter response rte ut no difference in remission rtes compred with shm control. Helth-Relted Outcomes of Nonphrmcologic Tretments (KQ 6) Direct evidence. With respect to ptient-reported helth-relted outcomes, we focused on qulity of life (vrious mesures) nd ility to function in dily life. One Tier 1 study compred ECT with comintion of ECT nd rtms nd found no differences etween groups in improvement on the Glol Assessment of Functioning scle (low strength of evidence). Indirect evidence. Two trils (oth in mixed MDD/ipolr popultions) ssessed generl helth sttus nd mentl nd physicl functioning (ll helth domins relted to qulity of life). In one fir tril, low rtms hd significntly greter improvement in helth sttus nd dily functioning thn shm, while this reltionship pproched sttisticl significnce when compring high rtms to shm (s mesured y the Glol Assessment of Functioning scle; low strength of evidence). In the other fir tril, VNS nd shm groups did not differ significntly in dily functioning (s mesured y the 36-item Medicl Outcomes Study Short Form [MOS SF-36]; low strength of evidence). No studies of psychotherpy were identified. Applicility For the limited mount nd low strength of evidence ville, the dt for Tier 1 (TRD) is generlly pplicle to TRD popultions. Popultions enrolled in these trils ppered representtive of our trget popultion. Studied interventions were comprle to those in routine use, though dose nd durtion of nonphrmclogic tretment often vried etween studies. Mesured outcomes on the whole reflected the most importnt clinicl outcomes for depression mesures, lthough reporting ws inconsistent; outcomes for the other key questions were much more restricted. Followup periods were generlly shorter thn desirle, ut most were sufficient to mesure n initil cutephse tretment response. Study settings were mixture of inptient nd outptient, ecuse ECT is generlly n inptient procedure nd the others re generlly outptient. Some evidence highlights the importnce of ptient cceptility of tretment s some ptients refuse prticulr interventions. An individulized lnce etween ptient s needs nd concerns must e tken into ccount during selection from rnge of nonphrmcologic nd phrmcologic ntidepressnt tretment options. The use of inconsistent definitions of TRD in the trils nd the sence of nlyses considering the effect of the numer of current tretment filures on outcomes hindered interprettion of dt, leding to our use of tiered system for nlyses. The evidence se comining dt for Tiers 1 3 on the whole produced findings tht were consistent with Tier 1 TRD dt nd lso pper pplicle to TRD popultions. Remining Issues This re of comprtive clinicl reserch is in its infncy. Key res for future reserch need primrily to ly more roust foundtions for n evidence se tht cn etter inform decisions for clinicins nd ptients. The field needs stndrd definition of TRD tht investigtors should use in their clinicl trils reserch. Comprison of ny of the potentil 13

14 interventions in the field, nonphrmcologic or otherwise, is hmpered y the vriility in TRD definitions. Although these definitions pper to e converging on single mening two or more tretment filures in the current episode very few studies of TRD hve pplied it. Progress in this re of reserch requires etter stndrdiztion of this concept, so tht future reviews of the evidence do not need to resort to differentiting, s we did, etween Tier 1 studies (i.e., TRD y this definition sed on two or more tretment filures) nd Tier 2 or 3 types of studies. The ltter do provide informtion tht helps illuminte likely impcts of these interventions on ptients with TRD, ut tht is not the sme thing s hving roust studies focused clerly on the ptient popultion of gretest interest. The chllenge will e to provide definition tht opertionlizes TRD to mke it fesile for clinicins while t the sme time successfully cpturing the complexity of tretment resistnce. More clinicl trils, s well s other possile study designs, tht compre nonphrmcologic interventions with other nonphrmcologic options nd with phrmcologic tretments re necessry to inform decisionmking in TRD. Clinicins, ptients, nd policymkers need dditionl relevnt dt to guide difficult tretment decisions out wht to do next: try nother mediction (nd should it e n ugmenttion, switch, or comintion strtegy?) or dd (or switch to) rtms, ECT, VNS, or psychotherpy? Also, given tht tretment options for mny TRD ptients include medictions, trils should directly compre nonphrmcologic interventions with ech other nd with phrmcologic tretments. The numer of tretment filures in the current episode should e delineted crefully. This informtion, more likely to e ccurte thn lifetime histories of filures, cn help investigtors determine whether the prticulr numer of filures, or reching prticulr numer of filures in current episode, cn help differentite etween nonphrmcologic tretment choices. For exmple, for ptients with two tretment filures in current episode, the outcomes my not differ etween cognitive therpy nd rtms; however, for ptients with different (higher or lower) numer of tretment filures in the current episode, one nonphrmcologic tretment my indeed e etter thn the other. Currently, we do not know wht the proper threshold is for selection of tretment. Clrifiction of the scientific sis for such decision would sustntilly improve decisionmking. Clrifying whether responses differ for TRD ptients with MDD compred with those with ipolr disorder will help guide future clinicl tril design. Our decision to include trils with ptient popultions including up to 20 percent with ipolr disorder (i.e., the mixed popultions noted erlier) ws guided y clinicl experience nd common sense ut not y dt. Testing to see whether outcomes differ etween the two groups cn yield informtion out inclusion criteri (should the mix e 0 percent, 10 percent, 20 percent, etc.?) tht my e useful to investigtors in designing TRD trils nd my e importnt to consider s potentil covrite in nlyses involving such mixes. Greter considertion should e given to the role tht the spectrum of depressive severity plys. Using finer grdtion of depressive severity thn investigtors now typiclly employ might identify whether prticulrly severe degrees of depression, most commonly understood currently s HAM-D17 20, my respond differently to the ville nonphrmcologic interventions thn do less severe levels of depression. These grdtions my led clinicins to etter understnding of severe depression nd its role in guiding tretment selection in TRD. Direct comprisons of tretment strtegies, holding consistent ny coexisting or concomitnt therpies, re impertive. Decisionmkers need to know whether outcomes with nonphrmcologic tretments re etter when such tretment ugments the current tretment, replces the current tretment, or replces the current tretment in comintion with nother tretment. When ongoing tretment is uncontrolled nd reflects vriety of tretments e.g., some ptients continue with typicl ntipsychotics, some with mood stilizers, some with no psychotropic medictions results of such studies re difficult, if not impossile, to interpret. 14

15 Consistent reporting of chnges in depressive severity, response rtes, nd remission rtes is crucil. To llow for etter comprisons of clinicl outcomes in this difficult-to-tret popultion, ll three mesures offer useful informtion for clinicins. Thus, for either clinicl trils or oservtionl studies, investigtors should ttempt to collect dt on ll three routinely. Appliction of consistent, ccepted protocols in trils is necessry. Mking sure tht ptients receive equivlent doses of different nonphrmcologic interventions is more difficult thn mking sure of this for phrmcologic interventions. Nevertheless, investigtors designing trils of nonphrmcologic therpies cn ttempt to do so y implementing stndrd ccepted protocols for their trils. Such dosing hd een difficult to control when tht protocol ws in the process of eing developed, s with rtms, ut given current tretment prmeters, this stndrdiztion is gol well worth trying to rech. More creful nd consistent ssessment of dverse events is required. Adverse event reporting is quite limited nd tends to cover only short time spn; wht reporting does exist is vrile nd inconsistent. Systemtic collection nd more consistent reporting of dt on hrms tht is, dverse events nd negtive side effects nd informtion out ttrition nd withdrwl would provide useful informtion to help lnce informtion now focused on clinicl enefits. Use of the CONSORT sttement (ville t: which guides proper reporting of study informtion (including the presenttion of dverse events), would strengthen reporting of oth hrms nd other clinicl tril findings; it would lso id in the criticl pprisl nd interprettion of ll study results. Further, more informtive ssessment of dverse events would require studies to e le to ssess longterm nd cumultive outcomes. Including key relevnt mesures nd sugroups in susequent reserch is desirle. As indicted y the review, nerly no evidence exists on how the effectiveness of nonphrmcologic tretments differs (or not) s function of symptom sutypes or for sugroups defined y sociodemogrphic chrcteristic (such s ge) or coexisting medicl conditions (e.g., post-stroke or postmyocrdil infrction depression; perintl depression). Also essentilly missing is informtion out helth-relted outcomes, especilly those reported y ptients, tht concern their qulity of life or levels of functionl impirment. Susequent studies should focus on employing known, relile, nd vlid mesures of ptient-reported outcomes, such s the MOS SF-36, the Qulity of Life Enjoyment nd Stisfction Questionnire (Q-LES-Q), nd the EQ-5D. Including comprisons of newer nonphrmcologic interventions will e importnt in future reserch. As new nonphrmcologic tretments re developed nd tested, investigtors should try to include them s potentil comprtors. At the time we strted this comprtive effectiveness review, clinicl tril dt on some of the developing nonphrmcologic interventions, such s mgnetic seizure therpy or deep rin stimultion, were insufficient (from the pulished literture) for us to try to include them. As the evidence ses grow to support the efficcy of such dditionl nonphrmcologic interventions, the newer strtegies should e included in comprtive effectiveness study designs. Conclusion Our review suggests tht comprtive clinicl reserch on nonphrmcologic interventions in TRD popultion is erly in its infncy, nd mny clinicl questions out efficcy nd effectiveness remin unnswered. Interprettion of the dt is sustntilly hindered y vrying definitions of TRD nd the pucity of relevnt studies. The gretest volume of evidence is for ECT nd rtms. However, even for the few comprisons of tretments tht re supported y some evidence, the strength of evidence is low for enefits, reflecting low confidence tht the evidence reflects the true effect nd indicting tht further reserch is likely to chnge our confidence in these findings. This finding of low strength is most notle in two cses: ECT nd rtms did not produce different clinicl outcomes in TRD, nd ECT produced etter outcomes thn phrmcotherpy. No trils directly compred the likelihood of mintining remission for nonphrmcologic interventions. The few trils ddressing dverse events, supopultions, sutypes, nd helth-relted outcomes 15

16 provided low or insufficient evidence of differences etween nonphrmcologic interventions. The most urgent next steps for reserch re to pply consistent definition of TRD, to conduct more hed-to-hed clinicl trils compring nonphrmcologic interventions with themselves nd with phrmcologic tretments, nd to delinete crefully the numer of tretment filures following tretment ttempt of dequte dose nd durtion in the current episode. For More Copies For more copies of Nonphrmcologic Interventions for Tretment-Resistnt Depression in Adults: Executive Summry No. 33 (AHRQ Pu. No. 11-EHC056-1), plese cll the AHRQ Cleringhouse t or emil hrqpus@hrq.gov. Full Report This executive summry is prt of the following document: Gynes BN, Lux L, Lloyd S, Hnsen RA, Grtlehner G, Thied P, Jons D, Brode S, Swinson Evns T, Crotty K, Viswnthn M, Lohr KN. Nonphrmcologic Interventions for Tretment- Resistnt Depression in Adults. Comprtive Effectiveness Review No. 33. (Prepred y RTI Interntionl-University of North Crolin [RTI-UNC] Evidence-Bsed Prctice Center under Contrct No I.) AHRQ Puliction No. 11-EHC056-EF. Rockville, MD: Agency for Helthcre Reserch nd Qulity. Septemer Aville t: AHRQ Pu. No. 11-EHC056-1 Septemer 2011