ARTICLES Effectiveness of Belt Positioning Booster Sets: An Updted Assessment AUTHORS: Kristy B. Arbogst, PhD, Jessic S. Jermkin, DSc, Michel J. Klln, MS, b nd Dennis R. Durbin, MD, MSCE,b Center for Injury Reserch nd Prevention, Children s Hospitl of Phildelphi, Phildelphi, Pennsylvni; nd b Center for Clinicl Epidemiology nd Biosttistics, University of Pennsylvni, Phildelphi, Pennsylvni KEY WORDS child pssenger sfety, booster sets, set belts ABBREVIATIONS AAP Americn Acdemy of Peditrics NHTSA Ntionl Highwy Trffic Sfety Administrtion BPB belt-positioning booster PCPS Prtners for Child Pssenger Sfety AIS Abbrevited Injury Scle OR odds rtio CI confidence intervl The results presented in this report re the interprettion solely of the Prtners for Child Pssenger Sfety reserch tem t the Children s Hospitl of Phildelphi nd re not necessrily the views of Stte Frm. www.peditrics.org/cgi/doi/10.1542/peds.2009-0908 doi:10.1542/peds.2009-0908 Accepted for publiction Jun 5, 2009 Address correspondence to Kristy B. Arbogst, PhD, Engineering, Center for Injury Reserch nd Prevention, Children s Hospitl of Phildelphi, 34th nd Civic Center Boulevrd, Suite 1150, Phildelphi, PA 19104. E-mil: rbogst@emil.chop.edu PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright 2009 by the Americn Acdemy of Peditrics FINANCIAL DISCLOSURE: The uthors hve indicted they hve no finncil reltionships relevnt to this rticle to disclose. WHAT S KNOWN ON THIS SUBJECT: Previous reserch demonstrting the benefits of booster sets over set belts for children is outdted. Becuse more children, prticulrly older children, re now restrined in booster sets, it is importnt to provide n updted ssessment of booster set effectiveness. WHAT THIS STUDY ADDS: This study reconfirms tht booster sets reduce the risk for injury in children ged 4 through 8 yers. BPB sets should continue to be recommended until t lest 8 yers of ge once child outgrows hrness-bsed restrint. bstrct OBJECTIVE: The objective of this study ws to provide n updted estimte of the effectiveness of belt-positioning booster (BPB) sets compred with set belts lone in reducing the risk for injury for children ged 4 to 8 yers. METHODS: Dt were collected from longitudinl study of children who were involved in crshes in 16 sttes nd the District of Columbi from December 1, 1998, to November 30, 2007, with dt collected vi insurnce clims records nd vlidted telephone survey. The study smple included children who were ged 4 to 8 yers, seted in the rer rows of the vehicle, nd restrined by either set belt or BPB set. Multivrible logistic regression ws used to determine the odds of injury for those in BPB sets versus those in set belts. Effects of crsh direction nd booster set type were lso explored. RESULTS: Complete interview dt were obtined on 7151 children in 6591 crshes representing n estimted 120 646 children in 116 503 crshes in the study popultion. The djusted reltive risk for injury to children in BPB sets compred with those in set belts ws 0.55. CONCLUSIONS: This study reconfirms previous reports tht BPB sets reduce the risk for injury in children ged 4 through 8 yers. On the bsis of these nlyses, prents, peditricins, nd helth eductors should continue to recommend s best prctice the use of BPB sets once child outgrows hrness-bsed child restrint until he or she is t lest 8 yers of ge. Peditrics 2009;124:1281 1286 PEDIATRICS Volume 124, Number 5, November 2009 1281
The Americn Acdemy of Peditrics (AAP) nd the Ntionl Highwy Trffic Sfety Administrtion (NHTSA) recommend the use of child restrint systems to protect children in crshes, including the use of child restrint systems with hrnesses for children from birth to t lest 4 yers of ge, followed by the use of belt-positioning booster (BPB) sets until they fit properly in the vehicle set belt. Despite these recommendtions, mny children begin using the vehicle belt premturely, 1 which puts them t n incresed risk for serious injuries in crsh. 2 4 Previous reserch demonstrted tht booster sets reduce the risk for injury to children ged 4 to 7 yers by 59% compred with similr-ged children in dult set belts by improving restrint geometry for children who re too smll for the vehicle set belt. 2 This previous nlysis, conducted on dt from 1998 to 2002, ws bsed primrily on children who were ged 4 nd 5 yers becuse of the usge prctices during tht period. In the time since tht reserch, pproprite restrint use mong children ged 4 through 8 yers hs incresed threefold. 5 This is ttributble, in prt, to mny sttes pssing upgrdes to their child restrint lws tht require booster sets for children who re older thn 4 yers. The upper ge limit of these booster lws vries by stte nd rnges from 6 to 8 yers. Pssge of these lws is ssocited with nerly 40% increse in child restrint use mong children through ge 7. 6 Becuse more children, prticulrly older children, re now ppropritely restrined in booster sets, we sought to provide n updted estimte of the effectiveness of BPB sets compred with the use of set belts lone in reducing the risk for injury for children ged 4 to 8 yers. The effectiveness of booster sets by impct direction nd booster set type ws lso explored. METHODS Dt Source Dt from the Prtners for Child Pssenger Sfety (PCPS) project from December 1, 1998, nd November 30, 2007, were used for these nlyses. PCPS consists of lrge-scle, child-specific crsh surveillnce system: insurnce clims from Stte Frm (Bloomington, IL) function s the source of subjects, with vlidted telephone survey nd on-site crsh investigtions serving s the primry sources of dt. 7 The prent/driverreported telephone survey served s the source of dt for the nlyses performed. Vehicles tht qulified for inclusion were Stte Frm insured, model yer 1990 or newer, nd involved in crsh with t lest 1 child occupnt who ws 15 yers of ge. Qulifying crshes were limited to those tht occurred in 15 sttes nd the District of Columbi, representing 3 lrge regions of the United Sttes (Est: NY, NJ [until November 2001], PA, DE, MD, VA, WV, NC, nd DC; Midwest: OH, MI, IN, nd IL; West: CA, NV, AZ, nd TX [strting June 2003]). Policyholders from qulifying crshes were contcted by the insurnce compny nd told tht they were eligible for motor vehicle sfety study. They were given very brief description of the study tht explined tht, with their consent, limited dt from their clim would be trnsferred electroniclly to reserchers t the Children s Hospitl of Phildelphi nd University of Pennsylvni. They were told to expect telephone cll from these reserchers for dditionl dt collection. Dt in this initil trnsfer included contct informtion for the insured, the ges nd genders of ll child occupnts, nd coded vrible describing the level of medicl tretment received by ll child occupnts (no tretment, physicin s office or emergency deprtment only, dmitted to the hospitl, or deth). Smpling nd Dt Collection A strtified cluster smple ws designed to select vehicles for the conduct of telephone survey with the driver of the vehicle nd prent(s) of the children in the smpled vehicle. Vehicles were strtified on the bsis of the initil medicl tretment received by child occupnts nd whether the vehicle ws drivble, nd probbility smple from ech tow sttus/medicl tretment strtum ws selected. When vehicle ws smpled, the cluster of ll child occupnts in tht vehicle ws included in the survey. Drivers of smpled vehicles in which t lest 1 child received medicl tretment were contcted by telephone, consented for telephone interview, nd screened vi n bbrevited survey to verify the presence of t lest 1 child occupnt with n injury. All vehicles with t lest 1 child who screened positive for injury nd 10% rndom smple of vehicles in which ll child occupnts screened negtive for injury were selected for full interview. (The 2.5% of smpled vehicles in which no children were treted were lso selected for full interview.) The full interview involved 30-minute telephone survey with the driver of the vehicle nd prents of the involved children. Often the driver nd the prent were the sme person; if not, then the interview ws conducted with the driver. On the bsis of n nlysis of dt for the period of this study, clim representtives correctly identified 97% of eligible vehicles, nd 80% of policyholders either consented for prticiption in this study or were not smpled for consent (the procedure to identify prticipnts who required consent chnged in June 2003). Of those who consented nd were smpled for n interview, 79% were successfully contcted nd screened for the full interview, representing n overll inclusion rte of 52% of eligible individuls. The 1282 ARBOGAST et l
ARTICLES included smple did not differ from known popultion vlues from Stte Frm clims with respect to geogrphic region, model yer of vehicle, tow sttus of the vehicle, nd ge of the child occupnt. TABLE 1 Child nd Driver Chrcteristics of the Study Smple Chrcteristic Overll, Weighted % (N 7151) Dt Anlysis Survey questions regrding injuries to children were designed to provide responses tht were clssified by body region nd severity on the bsis of the Abbrevited Injury Scle (AIS) score nd were previously vlidted to distinguish AIS 2 injuries from those less severe. 8 For the purposes of these nlyses, children were clssified s injured when they hd cliniclly significnt injury generlly corresponding to injuries with n AIS score of 2 (concussions nd more serious brin injuries, internl orgn injuries, spinl cord injuries, nd extremity frctures). Children who sustined only minor injuries generlly corresponding to n AIS score of 1, such s lcertions, contusions, nd brsions, were not considered injured for these nlyses. Current child restrint lws vry by stte but, collectively, incorporte children up through 8 yers of ge. Becuse of low usge rtes for children 9 yers, these nlyses were therefore restricted to rer-seted children who were ged 4 to 8 yers nd were restrined by BPB sets or set belts. The booster set restrined children were dditionlly ctegorized s those who used high-bck booster set or bckless booster set. Shield boosters were excluded from these nlyses becuse their use is not recommended for this ge group. To ccount for the potentil clustering of multiple children in single smpled vehicle nd the disproportionl probbility of selection of the study smple design, we used SAS-cllble SUDAAN: Softwre for the Sttisticl Anlysis of Correlted Dt 9.0 (Reserch Tringle Institute, Reserch Tringle Prk, NC) for the dt nlyses. Frequency distributions of severl child, vehicle, nd impct chrcteristics mong the smple were determined. Multivrible logistic regression ws used to determine the odds of injury for those in BPB sets versus those in set belts. Becuse the probbility of injury ws low in the smple ( 5%), these odds rtios (ORs) were interpreted s good pproximtions of risk rtios. Additionl nlyses exmined risk rtios between the 2 booster set types. RESULTS Between December 1, 1998, nd November 30, 2007, interviews were completed on 34 732 children in 21 943 crshes, representing 531 193 children in 346 485 crshes. From the overll PCPS smple, 7151 children in 6591 crshes met the inclusion criteri (rer seted, ged 4 8 yers, nd restrined in set belt or BPB set), BPB Set, Weighted % (N 1604) Set Belt, Weighted % (N 5547) Child chrcteristics Age, y 4 16.1 (1131) 30.8 (517) 9.7 (614) 5 21.5 (1454) 32.0 (495) 16.9 (959) 6 20.4 (1461) 19.0 (315) 21.0 (1146) 7 21.3 (1558) 13.6 (208) 24.7 (1350) 8 20.7 (1547) 4.7 (69) 27.7 (1478) 4 to 5 37.6 (2585) 62.7 (1012) 26.6 (1573) 6 to 8 62.4 (4566) 37.3 (592) 73.4 (3974) Weight, lb 40 9.5 (625) 18.6 (296) 5.5 (329) 40 59 56.4 (3829) 66.9 (1064) 51.8 (2765) 60 79 22.0 (1653) 11.6 (192) 26.5 (1461) 80 7.8 (601) 1.7 (27) 10.5 (574) Unknown 4.3 (443) 1.1 (25) 5.7 (418) Seting position Left outbord 41.7 (2916) 47.3 (747) 39.3 (2169) Center 14.1 (1148) 5.4 (96) 17.9 (1052) Right outbord 44.2 (3087) 47.3 (761) 42.8 (2326) Driver chrcteristics Age 25 y 5.0 (455) 5.5 (93) 4.7 (362).370 Mle gender 26.8 (1957) 24.2 (411) 27.9 (1546).032 Restrined 97.1 (6898) 97.9 (1561) 96.7 (5337).070 Prent 81.0 (5721) 87.1 (1394) 78.4 (4327) P vlues between BPB set nd set belt groups where pplicble. For ctegoricl (nondichotomous) vribles, the P vlues refer to the differences in the distribution of the vribles between the BPB set nd set belt groups. representing 120 646 children in 116 503 crshes. Overll, 70% of the children were restrined by set belt; the remining 30% were in BPBs. Tble 1 provides the distribution of ge, weight, gender, restrint use, nd seting position for the children nd driver/prent chrcteristics, including gender, restrint sttus, nd reltionship to the child by restrint type. For the overll smple, the children were pproximtely evenly divided cross the 4- to 8-yer ge rnge (20% 22% for ech yer of ge) with slightly fewer 4-yerold children (16%). Children who were restrined by the set belt were more likely to be older nd hevier thn those in boosters, lthough they remined within the recommended best prctice guidelines for booster set use. Children in both types of restrint were seted primrily in the outbord positions; however, lrger proportion of set-belted children occupied the center seting position P PEDIATRICS Volume 124, Number 5, November 2009 1283
TABLE 2 Crsh Chrcteristics of the Study Group Chrcteristic Overll, Weighted % (N 7151) BPB Set, Weighted % (N 1604) Set Belt, Weighted % (N 5547) Vehicle type.090 Pssenger cr 41.8 (3333) 41.3 (674) 42.1 (2659) Lrge vn 2.0 (161) 1.1 (27) 2.4 (134) Pickup truck 4.4 (296) 4.7 (73) 4.2 (223) SUV 25.0 (1568) 26.2 (404) 24.5 (1164) Minivn 26.7 (1793) 26.7 (426) 26.7 (1367) Model yer 1990 1997 37.3 (3275) 23.0 (438) 43.5 (2837) 1998 2001 38.5 (2547) 38.2 (597) 38.6 (1950) 2002 2008 24.2 (1329) 38.8 (569) 17.8 (760) Crsh severity.032 Any intrusion 7.7 (1454) 6.3 (300) 8.2 (1154) Towwy, no intrusion 27.4 (3001) 26.1 (639) 28.0 (2362) None 65.0 (2696) 67.6 (665) 63.8 (2031) Direction of initil impct.018 Front 46.6 (3312) 43.8 (695) 47.9 (2617) Ner side 9.4 (744) 11.9 (228) 8.3 (516) Fr side 11.6 (943) 12.0 (217) 11.4 (726) Rer 30.2 (1952) 30.6 (425) 30.0 (1527) Other/miscellneous/ unknown 2.2 (200) 1.8 (39) 2.4 (161) P vlues between BPB set nd set belt groups where pplicble. For ctegoricl (nondichotomous) vribles, the P vlues refer to the differences in the distribution of the vribles between the BPB set nd set belt groups. P compred with booster-seted children (18% vs 5%). Booster-seted children were more likely to be driven by prent (87% vs 78%; P.001), wheres setbelted children were driven more often by men (28% vs 24%; P.03). Tble 2 provides the distribution of vehicle type, model yer, crsh severity, nd impct type by restrint type. For children in both booster sets nd vehicle set belts, pssenger crs were the most common vehicle type, followed by minivns, SUVs, pickup trucks, nd lrge vns. Children who were restrined by the vehicle set belt were more likely to be in vehicles of older model yer nd more likely to be involved in crshes with intrusion present or requiring vehicle to be towed. In ddition, higher proportion of crshes for children in set belts were frontl impcts nd fewer were side impcts s compred with children in booster sets. The overll risk for AIS 2 nd greter injury ws 1.15% for ll 4- to 8-yerolds. Children in booster sets hd pproximtely hlf the injury risk s children in set belts (0.67% for children in BPB vs 1.36% for children in set belts). Tble 3 shows the undjusted nd djusted ORs of injury for BPB sets versus set belts. The djusted models ccount for child ge, weight, nd seting position; driver restrint nd reltionship to the child; crsh severity; model yer (if pplicble); direction of impct (if pplicble); nd crsh yer. After ccounting for potentil confounders, children who were ged 4 to 8 yers nd using BPB sets were 45% less likely to sustin injuries thn similrly ged children who were using the vehicle set belt (OR: 0.55 [95% confidence intervl (CI): 0.32 0.96]). Children in side impct crshes benefited the most from booster sets, showing reduction in injury risk of 68% for ner-side impcts nd 82% for fr-side impcts. There ws evidence tht children in booster sets in frontl impcts were lso t reduced risk for injury compred with those in set belts; however, we could not exclude the possibility of no difference. Children who were using booster sets in model yer 1998 nd newer vehicles hd the gretest risk reduction compred with children in belts (OR: 0.33 [95% CI: 0.18 0.61]). A totl of 61% (n 932) of the boosterseted children were restrined in high-bck booster sets; the remining 39% (n 672) were in bckless booster sets. Among children who were restrined in booster sets, we were not ble to detect difference in the risk for injury between the children in bckless versus high-bck boosters (OR: 0.84 [95% CI: 0.44 1.61]). Tbles 4 nd 5 show the distribution of AIS 2 injured body regions mong 4- TABLE 3 Undjusted nd Adjusted OR of Injury for BPB Sets versus Set Belts Prmeter Undjusted Model BPB vs Belts BPB vs Belts Adjusted Model OR 95% CI OR 95% CI Overll 0.49 0.31 0.78 0.55 0.32 0.96 Direction of initil impct Front 0.43 0.24 0.78 0.59 0.31 1.13 Ner side 0.25 0.14 0.46 0.32 0.13 0.79 Fr side b 0.31 0.12 0.77 0.18 0.06 0.49 Rer 0.59 0.28 1.24 0.97 0.35 2.74 Vehicle model yer 1990 1997 1.12 0.50 2.52 1.33 0.53 3.35 1998 2001 0.33 0.18 0.63 0.31 0.14 0.68 2002 2008 0.35 0.16 0.76 0.54 0.20 1.47 1998 nd lter 0.32 0.19 0.52 0.33 0.18 0.61 Adjusted for child ge, child weight, child seting position, driver restrint, whether driver prent, crsh severity, model yer (if pplicble), direction of impct (if pplicble), nd crsh yer. b Fr-side impcts include children sitting in the fr-side outbord, s well s center seting position. 1284 ARBOGAST et l
ARTICLES TABLE 4 Body Region Distribution of Injuries Among 4- to 8-Yer-Olds by Restrint Type Injured Body Region Overll, Weighted % (N 7151) BPB Set, Weighted % (N 1604) b to 8-yer-old children by restrint type. Hed injuries were the most prevlent mong ll injured children, regrdless of restrint type, ccounting for 65% of injuries. Among injured children in booster sets, fce nd lower extremity injuries were the next most common t 9% nd 8%, respectively. Injured children in set belts sustined injuries to the bdomen nd fce t 12% nd 9%, respectively. For both types of BPB sets, hed injuries were the most common injury sustined, representing 59% nd 73% of injuries for bckless nd high-bck BPB sets, respectively. Hed injuries were followed by fce injuries for bckless BPB sets nd bdominl injuries for highbck BPB sets. Of those with n AIS 2 injury, 9%, 7%, nd 14% of children in set belts, high-bck booster sets, nd bckless booster sets, respectively, sustined n AIS 2 injury to 1 body region. There ws not significnt difference between the percentge of children with multiple injuries between those in high-bck versus bckless booster sets (P.26). DISCUSSION Set Belt, Weighted % (N 5547) c Hed 65.4 (471) 67.4 (78) 65.0 (393) Fce 8.7 (110) 9.4 (14) 8.5 (96) Chest 2.6 (28) 4.0 (8) 2.3 (20) Abdomen 10.7 (87) 5.1 (5) 12.0 (82) Neck/spine 1.5 (14) 1.1 (3) 1.6 (11) Upper extremity 6.6 (89) 5.4 (15) 6.9 (74) Lower extremity 4.5 (66) 7.6 (18) 3.8 (48) Totl n 722 with injury (1.15%). b Totl n 118 with injury (0.67%). c Totl n 604 with injury (1.36%). TABLE 5 Body Region Distribution of Injuries Among Booster-Seted Children by Booster Set Type Injured Body Region Bckless Booster Set, Weighted % (N 672) High-Bck Booster Set, Weighted % (N 932) b Hed 58.9 (42) 72.8 (36) Fce 17.8 (7) 4.1 (7) Chest 2.8 (3) 4.7 (5) Abdomen 0.0 (0) 8.3 (5) Neck/spine 2.8 (3) 0.0 (0) Upper extremity 4.7 (5) 5.9 (10) Lower extremity 13.1 (11) 4.1 (7) Totl n 57 with injury (0.63%). b Totl n 61 with injury (0.70%). This study used the most recent dt vilble to reexmine the effectiveness of BPB sets nd extends previous reports tht booster sets reduce the risk for injury in children by studying greter percentge of older children; 37% of the study smple who used booster sets were 6 to 8 yers of ge. The nlysis confirmed tht children who were ged 4 to 8 nd used BPB sets were 45% less likely to sustin injuries thn similrly ged children who used the vehicle set belt when considering ll crsh directions nd vehicle model yers. Children in side impcts derived the lrgest reltive protection from booster sets, with reduction in risk of 68% nd 82% for ner-side nd frside crshes, respectively. Reductions in injury risk in side impct crshes for children who were restrined in BPB sets were previously reported, 3 nd the nlyses on this lrger dt set suggest n even lrger protective effect of boosters in this impct direction. Side impct crshes often hve substntil frontl component. 9 The shoulder portion of the set belt my hve better fit on the child s shoulder when the child is in booster set nd therefore provide better protection thn shoulder belt tht fits poorly in the bsence of booster set. The lrgest reltive benefit ws relized for children who were seted fr side to the crsh, for which the risk for torso rollout from the shoulder belt is gretest. Of interest, lthough the OR suggested injury risk reduction for children in boosters in frontl impcts compred with those in set belts, the results did not rech sttisticl significnce for this dt set. One reson my be becuse of the chnging lndscpe of restrint prctices. Our previous nlysis reported n injury risk of 1.95% for 4- to 7-yer-old children in belts compred with 0.77% for boosters. 2 In this nlysis, we report n injury risk of 1.36% for children in belts compred with 0.67% for children in boosters. The reduction in injury risk for children in belts is likely ttributble to 2 resons. First, the previous study included children in both the front nd rer rows, wheres this study ws limited to those in the rer row. Front seting is ssocited with n incresed risk for injury, nd the proportion of set belt restrined children in the front row ws higher thn those in booster sets. Second, becuse more of the children in the 4- to 8-yer ge rnge use boosters, fewer of the smllest children (ie, those most susceptible to injury from poor belt fit) re using belts; therefore, s the popultion PEDIATRICS Volume 124, Number 5, November 2009 1285
of set belt users shifts towrd the older children, they re more likely to fit better in the vehicle set belt nd their overll injury risk reduced. Importntly, these results suggest tht the effectiveness of booster sets does not vry by the type of booster set: bckless or high-bck. Bckless booster sets re less costly nd often more cceptble to older children becuse of the bsence of bck tht mkes them look like toddler child restrint. These results give confidence to prents nd helth eductors tht choosing this type of restrint for their child does not represent compromise in sfety. Hed injuries remined the most commonly injured body region for ll of the restrined children in this study; however, bdominl injuries were the second most common injuries for belted children s result of set-belt syndrome injuries. Children who were restrined by booster sets sustined injuries to the fce nd lower extremity, with notble bsence of bdominl injuries. This reserch ws conducted on crshes tht involved Stte Frm policyholders only. Stte Frm is the lrgest insurer of utomobiles in the United Sttes, with more thn 38 million vehicles covered; therefore, its policyholders re likely representtive of the insured public in the United REFERENCES 1. Winston FK, Durbin DR, Klln MJ, Moll EK. The dnger of premture grdution to set belts for young children. Peditrics. 2000;105(6): 1179 1183 2. Durbin DR, Elliott MR, Winston FK. Beltpositioning booster sets nd reduction in risk of injury mong children in vehicle crshes. JAMA. 2003;289(21):2835 2840 3. Arbogst KB, Klln MJ, Durbin DR. Effectiveness of high bck nd bckless beltpositioning booster sets in side impct crshes. Annu Proc Assoc Adv Automot Med. 2005;49:201 213 4. Elliott MR, Klln MJ, Durbin DR, Winston FK. Sttes. This study obtined nerly ll of its dt vi telephone interview with the driver/prent of the child nd is therefore subject to potentil misclssifiction. Ongoing comprison of survey dt with crsh investigtion dt hs reveled high degree of greement between the 2 sources. In previous published nlyses of PCPS dt regrding risk for injury to children in compct extended-cb pickup trucks 10 nd the effectiveness of booster sets, 2 sensitivity nlyses tht were conducted to quntify the potentil impct of misclssifiction bis indicted tht n implusible mount (40% 50% misclssifiction of restrint use) of misclssifiction would be required for results of our nlyses to lose sttisticl significnce. Although some degree of misclssifiction likely exists in prent-reported dt, we believe tht it is of mgnitude tht would not lter the conclusions of this report. Our study smple represents the entire spectrum of crshes reported to n insurnce compny, from those with minor vehicle dmge to those with loss of life. It must be noted, however, tht given this distribution of crshes, we re looking lmost exclusively t nonftl injuries. CONCLUSIONS This study reconfirms previous reports tht BPB sets reduce the risk for injury in children 4 through 8 yers of ge by studying greter percentge of children ged 6 to 8 yers thn previous studies. After djustment for potentil confounders, children who were ged 4 to 8 nd using BPB sets were 45% less likely to sustin injuries thn similrly ged children who were using the vehicle set belt. Among children who were restrined in BPB sets, there ws no evidence of difference in the performnce of bckless versus high-bck boosters. On the bsis of these nlyses, prents, peditricins, nd helth eductors should continue to recommend s best prctice the use of BPB sets once child outgrows hrnessbsed child restrint until he or she is t lest 8 yers of ge. ACKNOWLEDGMENTS This work would not hve been possible without the commitment nd finncil support of Stte Frm for the cretion nd ongoing mintennce of the PCPS progrm, n ongoing collbortion mong Stte Frm, the Center for Injury Reserch nd Prevention t the Children s Hospitl of Phildelphi, nd the University of Pennsylvni. PCPS serves s the source of dt for the nlyses conducted herein. The Injury Center reserch tem cknowledges the mny Stte Frm customers who served s the subjects in PCPS. Effectiveness of child sfety sets vs set belts in reducing risk of deth in children in pssenger vehicle crshes. Arch Peditr Adolesc Med. 2006;160(6):617 621 5. PCPS, Prtners for Child Pssenger Sfety Fct nd Trend Report. 2008. Avilble t: www.chop.edu/injury. Accessed Mrch 1, 2009 6. Winston FK, Klln MJ, Elliott MR, Durbin DR. Effect of booster set lws on pproprite restrint use by children 4 to 7 yers old involved in crshes. Arch Peditr Adolesc Med. 2007;161(3):270 275 7. Durbin DR, Bhti E, Holmes J, et l. Prtners for child pssenger sfety: unique childspecific crsh surveillnce system. Accid Anl Prev. 2001;33(3):407 412 8. Durbin DR, Winston FK, Applegte SM, Moll EK, Holmes JH. Development nd vlidtion of the injury severity ssessment survey/prent report: newinjuryseverityssessmentsurvey. Arch Peditr Adolesc Med. 1999;153(4):404 408 9. Arbogst KB, Ghti Y, Menon RA, Tylko S, Tmborr N, Morgn R. Field Investigtion of child restrints in side impct crshes. Trffic Inj Prev. 2005;6(4):351 360 10. Winston FK, Klln MJ, Elliott MR, Menon RA, Durbin DR. Risk of injury to child pssengers in compct extended-cb pickup trucks. JAMA. 2002;287(9):1147 1152 1286 ARBOGAST et l