The Florida trauma system: assessment of a statewide data base

Copyright

Injury Vol. 27, No. 3, pp. 205-208, 1996 0 1996 Elsevier Science Ltd. Ali rights reserved Printed in Great Britain 00.X-1383/96

$lS.OO+O.OO

ELSEVIER

0020-1383(95)00193-x

The Florida trauma system: statewide data base

assessment

of a

H. Rodenberg Division of Emergency Medicine, University

of Florida College of Medicine, Gainesville, Florida, USA

The State of Florida mainfains a statewide trauma registry. i-his paper examines the data provided to the registryby four Level I, four Level II, and four non-designated traumacare facilities within the sfafe for complefion. There were18961pafienfrecords in the registry between I ]uly 1991 and 30]une 1992 which were reviewed. Completed records were defined as confaining appropriate notations for age, mechanism of inju y, Glasgow Coma Scale, systolic blood pressure, repirafo y rate, and Injury Severity Score. The mean completion rate in Level Ifacilities was 22.75 per ten f, in Level II facilities if was 79.15 per cent and in non-designated cenfres if was 3 1.95 per cent. The deficits in completion may lead to errors in calculating trauma sysfem efficacy,outcomes, and costs. Assigning and enforcing responsibility for completion of frauma registry data within individual facilities may aid in the resolution of fhese issues. Copyright 0 1996 Elsevier Science Ltd. All rights reserved.

Injury, Vol. 27, No. 3, 205-208, 1996

Introduction Trauma registries are intended to create data bases on injured patients cared for within a given institution, association,region, or governmental jurisdiction. The data collected by trauma registries is the key to the accurate determination of outcome statistics (both morbidity and mortality) and the rates of over- and undertriage to designated trauma centres and other receiving facilities. This information is critical to health plannersand managers as they devise new trauma systems, improve existing programmes, and determine the medical, economic, and medicolegal impacts of trauma careIe9. Comprehensive and accurate collection of patients’ data is crucial to the successfulfunctioning of a trauma registry. The Division of Emergency Medical Services (EMS) of the Department of Health and Rehabilitative Services of the State of Florida instituted a statewide trauma registry in 1989. State rule requiresthat eachtrauma patient identified by prehospital care providers or emergency department personnel must be the subject of a trauma registry form (HRS 1728). The data sheet is to be initiated by the first health care provider to evaluate the patient, and is to follow the patient upon transfer to an emergency department, another health care facility, or an inpatient unit. Upon discharge, the form is to be completed by hospital personnel and sent to the state EMS office for entry into

the trauma registry. State rule indicates that the data sheet is to be completed on every patient who suffers from injury, no matter what the severity of the injury or the destination facility might be“‘. The aim was to evaluate the completenessof the data in the state trauma registry provided by institutions representing both designated trauma centres and community facilities.

Methods All State of Florida trauma registry patient records for the period 1 July 1991 to 30 June1992 were obtained from the four state-designated Level I trauma centres, four randomly selected Level II facilities, and four non-designated hospitals in north central Florida. The latter group comprised three full-service community hospitals and one full-service academicmedical centre. Patients who arrived at one of the study institutions after an interfacility transport were excluded from evaluation. Records were provided in electronic form. Eachtrauma registry record was scannedfor patient age, mechanism of injury (blunt versus penetrating), initial blood pressure,initial Glasgow Coma Scale(GCS) score”, initial respiratory rate, and Injury Severity Score (ISS)1z,13. These six parameters were chosen for study as they represent the minimum data set required to be able to calculate the Revised Trauma Score (RTS) and determine the probability of survival using TRISS methodoA patient record was defined as complete if it logy 1,14,15. included appropriate notations for all these parameters. Trauma registry records were reviewed for completeness by the individual institution and by groups based on trauma centre status. Records were further examined by data item to determine any highlights in data deficiencies.

Results The samplepopulation comprised 18 961 trauma registry records. Of these records, 6761 were from Level I trauma centres, 8856 were from Level II, and 3344 records originated from full-service community hospitals. After review, 433 I recordswere found to be complete for the six data points studied. The overall completion rate was 22.8 per cent. The percent of completed records in Level I centres was 22.75

206

Injury:

International

Journal of the Care of the Injured

Vol. 2 7, No. 3, 1996

Table I. Trauma recordcompletionrate Facility

code

Level I centres I-A I-B I-C I-D Mean = 22.75% Level II centres II-A II-B I I-C II-D Mean = 19.75% Non-designated ND-A ND-B ND-3 ND-4 Mean = 31.95%

Total no. of cases

No. completed

1904 1142 2335 1380 Level I centres

completed

within

completed

2922 547 3302 2085 within Level

facility

% total complete

705 308 474 92

37.0 27.0 20.3 6.7

16.3 7.1 10.9 2.1

482 219 531 213

16.5 40.0 12.3 10.2

11.1 5.1 16.1 4.9

82 169 46 110

71.3 18.9 6.4 31.2

22.7 3.9 1.1 2.5

II centres

facilities

completed

within

1377 894 720 353 non-designated

facilities

Reasonsfor non-completion of trauma records (numberreviewed, 18961) TableII.

Parameter

% completion,

not recorded

Patient’s age Initial respiratory rate Initial systolic blood pressure lntitial GCS’ Blunt vs penetrating trauma Injury Severity Score’

No. of records

% total records*

131 6386 5436 3176 10087 9935

0.7 33.7 28.7 16.8 53.2 52.4

Total number records completed = 4331; Total % of records completed = 22.8%. *Recording of GCS ~3. ‘Recording ISS=Oor 99. *Total > 100% due to multiple omissions on some records.

per cent, in Level II centres it was 19.75 per cent and in community facilities it was 31.95 per cent. Further analysis of individual items revealed that 6386 records had no notation of respiratory rate, 5436 had no indication of initial systolic blood pressure,3176 did not record a GCS score, 10 087 did not indicate whether the patient suffered from blunt or penetrating injury, 9935 did not yield a valid ISS score,and 131 did not list the patient’s age. Individual records exhibited many deficits in completenessof data. These results are summarizedin Taables I and II.

Discussion Reviewing the database establishedby the Florida EMS Office reveals a number of points of interest, First and foremost is the deficit exhibited by designated trauma centres in providing complete data. In Florida, designated trauma centres are required to have a designated trauma, care coordinator, one of whose functions is to collect trauma registry data for both the local institution and the state”. However, in Level I trauma centresthe completion rate for the six items studied ranged from 6.7 per cent to 37.0 per cent, in Level II centres,from 10.2 per cent to 40.1 per cent and in non-designated centresfrom 6.4 per cent to 71.3 per cent, of ail charts. It is also worth recalling that only six items on the trauma registry form were studied; it

is likely t’hat the completion rate for all items requested on the form is even less. Certainly trauma centre coordinators are busy with many duties, and it might be that their efforts were focused on compiling data on the more severely injured patients or those who suffered complications. It cannot be determined from this study if there was a tendency for charts to be completed in these groups; since many patients did not have enough data to complete an RTS, and an ISS was often omitted from the form, it is difficult to know what the true distribution of patients by severity of injury was for each institution or group. It may be that trauma centre coordinators are not at fault, especially if they designate another individual to perform trauma registry data collection, the forms are completed by medical records personnel, or thesefunctions are conducted by a local or regional trauma overseeing agency. Deficits in completing the initial component of the form may be attributable to EMS service omissions;however, these values are to be confirmed by the trauma coordinator reviewing the EMS run report and filled in as required. Ultimately, it is the responsibility of the coordinator to track data at their own institution, and insure that accurate data is being transmitted to the state EMS office. The state office, in turn, needs to conduct periodic audits of the trauma registry and notify institutions when completion rates on data collection forms fall below acceptablestandards. Another item of interest is the apparent lack of familiarity by those individuals initiating or completing trauma registry records with scoring systems in common use in prehospital care, emergency medicine, and the study of trauma. It was surprising to discover that over 3100 patients had a GCS of lessthan 3, as a value of 3 is the minimum permittedl’. Similarly, over WOO records could not be analyseddue to ISSscoresof 0 or 99, both of which are impossibilities’3,14. There are several other potential sourcesof error. Given the low rate of successfulcompletion of trauma registry records, it seemslikely that many victims of injuries may not have been entered into the registry at all. The impact of these omissions,whether significant or negligible, cannot be estimated. It is possiblethat appropriate data may have occasionally been entered into inappropriate placeson the

Rodenberg:

The Florida

trauma

system

207

trauma registry form. This error alsolimits the utility of the data. Statistical analysis beyond simple calculation of completion rates of trauma registry forms could not be performed with confidence. It was impossible to ascertain that data was collected in a uniform manner in all study institutions, or that the population of injured patients on whom data were to be collected were alike. Also not reviewed were the presenceor effect of internal procedures within each’institution to increasethe yield of valid data. Given these concerns, the determination of whether statistically significant differences existed between institutions for the completion rate of trauma registry forms was impossible. The lack of additional analysis may impact upon the authority of the data presentedherein; but it is in itself a reflection of the fact that no adequate, standardized system of data collection exists within the institutions reviewed. The results of this study indicate that less than onequarter of the trauma registry data provided by the study institutions is sufficiently complete to allow calculation of basic mortality statistics and probabilities of survival. If adequate data cannot be collected, the trauma system is significantly damaged. Inadequate data may lead to inadequatefinancing of trauma services,underestimation of the impact of trauma on a local areaor region, and an inability to compare institutions or systems to determine the optimal meansto care for the injured patient. Conversely, it may also result in allocation of excessive resourcesfor trauma care where funds and personnel might be better used for other things. This study reviews only the trauma registry of one state, and focuses only upon selected institutions that contribute to the registry. While the experience of other states,regions, or institutions is unknown, it is anticipated that the problems which plague trauma registries in Florida may occur elsewhere. Adequate data collection may require individuals to be funded specifically for that purpose, as often occurs in the course of a clinical study. The current economic climate makes this increasingly unlikely, although the need for cost effectiveness and continuous quality improvement programmes may drive additional efforts.

Table III. Traumapatient outcomes:z statistic

Outcomes measurement: useful?

Noteworthy deficienciesexist within the State of Florida trauma registry. Calculation of the z statistic demonstrates that all Level I insitutions, three of the Level II, and three of the undesignated facilities studied exhibit a higher than expected death rate for injured victims within their institutions. Expansion of the data base to include complete information on all injured victims seenby health care facilities is required to verify these findings.

is the database

Any study which attempts to draw conclusions from incomplete or unmatched databases treads on tenuous ground. However, a casecanbe madethat while analysisof incomplete data will not lead to definitive conclusions, it can serve asan indicator of trends if the study population is thought to be representative and any suggestionsasto the meaning of the data are restricted to settings where variables are minimized. These data were subjected to additional demographic evaluation as a ‘trial balloon’ to detect possible trends in patient outcomes between trauma centres offering different levels of care. Patient demographicsbetween the Level I, Level II, and non-designated facilities varied widely, SO no valid comparisonof facility by patient population, RTS, or TRISS methodology could be performed. Champion and colleagues described the z statistic as a meansto expressthe actual number of trauma victims who survive their injury as compared with the number of patients who would be statistically predicted to survive

Trauma

Level/ID

Number

Code

of patients

z statistic

I-A I-B I-C I-D

705 308 474 92

5.31 1.44 4.07 2.24

II-A II-B I I-C II-D

482 219 531 213

2.77 1.62 - 4.62 0.99

FSC-A FSC-B FSC-C FSC-D

982 169 46 110

1.04 1.50 -0.32 -1.74

FSC= Full-service designation.

community

hospital,

no trauma

centre

based on TRISS methodology. While the possible values of z are infinite, a z value of lessthan I indicates that the trauma death rates in a given population are lower than would be expected for the calculated probabilities of survival of the victims. A value greater than 1 signifiesthat death rates are greater than predicted. While the z statistic cannot be used to compare facilities due to the variances in patient population, it can serve asa valuable internal check on the results of trauma care provided within an institution where patterns of care are well established,variables held to a minimum, and no deliberate biasin completion of data forms is suspected15,17. Analysis of the z statistic within each institution reveals that only three out of the 12 studied institutions had a result less than I (TubfeIfl). One of these was a Level II centre and the others were non-designated facilities. That this result might be the caseis only mildly surprising,given previous literature which describesthe ability of Level II Centres to perform at Level I standards18r1g. The variances in the databasedo not allow any definitive conclusion to be drawn, but the trend for positive z statistics to reside outside of Level I Centres may act as a spur for further investigation.

Conclusions

Acknowledgements The author wishes to acknowledge his great appreciation to Edith Orsini and Carol Gormley of the North Central Florida Health Planning Council for their assistancewith this work.

References 1 Eposito TJ, Nania J and Maier RV. State trauma system evaluation: a unique and comprehensiveapproach.Ann berg

Med

1992;

21: 35 1.

208

Injury:

International

2 American College of Emergency Physicians. Trauma care systems quality improvement guidelines. Ann Emerg Med 1992; 21: 736. 3 Champion HR, Copes WS, Sacco WJ et al. The major trauma outcome study: establishing national norms for trauma care. ] Trauma 1990; 30: 1356. 4 West JG, Cales RH and Gazzaniga AB. Impact of regionalization: the Orange County experience. Arch Surg 1983;

812. 7 Davis JW, Hoyt DB, McArdle MS et al. An analysis of errors causing morbidity and mortality in a trauma system: a guide for quality improvement. J Trauma 1992; 32: 660. 8 Smith JS, Martin LF and Young WW. Do trauma centers improve outcome over non-trauma centers: the evaluation of regional trauma care using discharge abstract data and patient management categories. J Trauma 1990;30: 1533. 9 Cottington EM, Young JC, Shufflebarger CM et al. The ability of physiologic status, injury site, and injury mechanism in identifying patients with major trauma. 1 Trauma

1988;28:305. 10 Florida Administrative

Code, Chapter lOD-66, Section 103. Effective December 10,1992; amended November 30,1993. I I Pons PT. Head trauma. In: Rosen P, Barkin RM, Braen GR et

of the Care of the Injured

Vol. 27, No. 3, 1996

al., Emergency Medicine: Concepts and Clinical Practice. St Louis: Mosby-Year Book, 3rd ed, 1991, p 343 12 Champion HR, Sacco WJ, Copes WS et al. A revision of the trauma score. ] Trauma 1989; 29: 623. 13 Copes WS, Champion HR, Sacco WJ et al. The injury severity score revisited. J Trauma 1988; 28: 69. 14 Greenspan L, McLellan BA and Greig H. Abbreviated injury scale and injury severity score: a scoring chart. ] Trauma

118: 740. 5 Maul1 KI, Schwab CW, McHenry SD et al. Trauma center verification. ] Trauma 1986; 26: 521. 6 Schackford SR, Hollingworth-Fridlund P, Cooper GF et al. The effect of regionalization upon the quality of trauma care as assessed by concurrent audit before and after institution of a trauma system: a preliminary report. ] Trauma 1986; 26:

Journal

1985;25:60. I5 Boyd CR, Tolson MA and Copes WS. Evaluating trauma care: the TRISS method. ] Trauma 1987; 27: 370. 16 Florida Department of Health and Rehabilitative services: State-Approved Trauma Center and State-Approved Pediatric Trauma Referral Center Approved Standards (HRSP 150-9). Tallahassee, State of Florida, 1991. 17 Cottington EM, Shufflebarger CM and Townsend R. The power of the z statistic: implications for trauma research and quality assurance review. ] Trauma 1989; 29: 1500. 18 Thompson CT, Bickell WH, Siemens RA et al. Community hospital level II trauma center outcomes. J Tratrma 1992; 32:

336. 19 Zulick LC, Dietz PA and Brooks K. Trauma experience of a rural hospital. Arch Swg 1991; 126: 1427. Paper accepted

24 October 1995.

Requests for reprints should be addressed to: Dr Howard Rodenberg FACEP, c/o Medical Rescue International, Johannesburg, Republic of South Africa. MD,