- Email: [email protected]
Aspects of neurosurgical assessment using the Glasgow Coma Scale
ad Critical Can Nwring (1992) Longman Gmup UK Ltd 1992
hhsti 0
8,9&99
Aspects of neurosurgical assessment using the Glasgow Coma Scale Anne Ellis and Stephen J. Cavanagh
The Glasgow Coma Scale (GCS) has become a cornerstone of the neurological/ surgical assessment of patients used by both nursing and medical staff. Since its development in the 1970s it has been used in a variety of clinical situations to monitor changes in a number of key neurological functions, including level of consciousness, pupil reaction and limb movement. During this time, however, ther have been suggestions that there are problems with some of the measurement principles underlying its use, whch in part has stimulated the development of otht neuro-assessment tools. Irrespective of measurement device, there is always the possibility of error or incorrect assessment. In the field of neurosurgery, as with other high dependency environments, a patient’s condition can change rapidly. Additionally, there is the association of certain assessment responses with nursing and medical interventions. Thus, accuracy in all aspects of assessment and recording is paramount. Despite the growing body of literature surrounding the GCS, little is known about the pattern of errors made by nursing staff using the GCS to assess neurosurgical patients. This study compared the assessment finding of Registered General Nurses (RGNs), Enrolled Nurses and Student Nurses after viewing videotaped neuro-assessments of patients in a high dependency unit. The criterion for judging the accuracy of subject’s assessments was established by a panel of experts. As expected, RGNs had the highest proportion of correct assessments and students the least. Subjects were identified as having difficulty in determining the relative amounts of weakness that a patient exhibited, and in correctly distinguishing between flexion and extension.
INTRODUCTION The practical needs of comatose and neurologitally impaired patients are varied. As a means to Anne Ellis BSc(Hons) RGN RM, Senior Clinical Nurse Manager, Midland Centre for Neurosurgery and Neurology, Stephen J. Cavanagh RGN MS PhD, Division of Health and Nursing Studies Wolverhampton Polytechnic (Requests for offprints to SC) Manuscript accepted 17 February 1992
94
standardise the vocabulary and assessment of patients following head injury, the Glasgow Coma Scale (GCS) was devised by Teasdale & Jennett in the 1970s. The principle underlying this scale is that it should be simple to use by non-medical health providers and more experienced neurological/surgical staff. The scale has been a focus for considerable medical research into its application as a prognostic indicator following head injury (Choi, Ward SCBecker, 1983), and the inter-rater reliability of staff
INTENSIVEANDCRlTlCALCARENURSlNG
using it (Braakman, Avezatt, Maas, Roe1 8c Schouten, 1977). There remains, however, relatively little nursing research related to the GCS. Yet the studies which have been performed address important nursing care issues, for example, Grant, Kinney & Guzzetta (1990) proposed that the GCS could be used as a means of validating nursing diagnoses, and Neatherlin & Brillhart (1988) used the GCS as an assessment tool with patients following cardio-pulmonary resuscitation. It is recognised that it is not possible to measure directly a concept such as level of consciousness with a single assessment tool. The underlying premise of the GCS is that continuous monitoring under ‘standard conditions’ will lead to a determination of a trend in a patient’s condition which can be more readily interpreted. The majority of this recording is currently undertaken on 24 h basis by either Kegistered General Nurses (KGNs), Enrolled Nurses (ENS) or Student Nurses (SNs) with appropriate supervision. While there has been some standardisation in the assessment and recording of assessment data using the GCS, a neurological/surgical patient can present severe challenges to the clinical judgement of the most experienced practitioner. Misinterpretation of clinical signs is possible as are errors in recording. Many ‘difhcult’cases will be resolved by either a consensus approach or by seeking additional assistance, nevertheless assessment errors will occur. In view of the knowledge that the clinical condition of neurological/ surgical patients can exchange extremely rapidly and that clearly errors can be made, it is perhaps surprising that little is known about the nature of assessment errors made by nursing staff using the GCS. This information will be useful to educators and clinicians alike in identifying improvements in the GCS, the teaching of assessment skills, and adopting the scale in nursing practice.
Purpose The purpose of this study was to investigate the pattern of errors made by RGNs, ENS and SNs when assessing neurosurgical patients.
95
METHODS Sample A random sample of 27 registered and 16 enrolled nurses employed in a regional neurological/neurosurgical centre were used as subjects in the study. In addition, a convenience sample of 22 third-year-student nurses with no previous neuroscience experience were chosen as comparison subjects. 77% of RGNs and 81% of ENS had more than 1 full-time year of neurosurgical experience. In addition, 32% of the RGNs possessed the English National Board (ENB) 148 and 10% the ENB 100 certificates. 12% of ENS had successfully completed the ENB 153 course.
Procedure Before commencing this study permission was obtained from the Ethical Committee, and a letter of consent was completed by each patient or relative involved. A video was made of the assessment of 19 patients, of these 12 were chosen as being representative of a variety of neurological responses, i.e. alert and orientated, no limb deficits, confused, mild limb deficits, severe deficits, flexing or extending to pain, or no response at all. An attempt was made to have patients in the same position (on their backs in bed) and for the nurses carrying out assessments to perform them in the same order. Stimuli were applied in an identical manner (nailbed pressure) to obtain the maximum response. This was not always possible as some patients were not able to remain in a recumbent position. There were also technical difficulties with filming, in particular with the recording of pupil responses; this item was excluded from the study. Of the 12 patients recruited to this study, 2 had no deficits, 5 had left-sided v-roakness, 4 right-sided weakness, and 1 was unresponsive to any stimulation. A panel of experts consisting of 3 registered nurses and 1 senior registrar (all with considerable experience in neurosciences) was assembled to view the video and determine an appropriate assessment response to be recorded on the GCS.
96
INTENSlVEANDCRlTlCALCARENURSlNG
Their agreed ‘standard’ was used as the criterion to rate the responses of the nursing subjects when they viewed the video.
Non-statistically
significant
findings
The following assessment areas produced significant findings at the the 0.5 level.
non-
Prior to the study all subjects were given: 1. an introduction and explanation of the study 2. a demographic questionnaire, including years in neuroscience and any relevant courses 3. a diagram and written explanation of terms (e.g. localise to pain, flexion, extension) 4. a GCS chart for recording results observed on each of the 12 patients. Subjects observed the video in groups of approximately 6. They were given time to read the instructions and asked not to confer. The video was stopped between each patient assessment
to allow
responses.
the
participants
to record
Eye opening
Correct
Enrolled
256
142
220
68
49
46
Incorrect
nurses
Students
V’ (Cramer) = 0.006; X’ = 4.71; df = 2; p = 0.094 Verbal
Correct
Staff nurses
Enrolled
283
155
214
41
37
98
Incorrect
nurses
Students
V’ (Cramer) = 0.007; X2 = 5.70; df = 2; p = 0.057 Left arm
their
It took 45 min to assess all patients.
Staff nurses
Staff nurses Correct Incorrect V2 (Cramer)
171 153 = 0.0054;
Enrolled nurses
Students
86
121
106
143
X’ = 4.18; df = 2; p = 0.1236.
RESULTS AND INTERPRETATION
Left leg Staff nurses
Enrolled nurses
Students
A tally was made of subjects’ assessments based on the video, and these were judged against the criteria agreed by the panel of experts. For the purposes of analysis, an individual’s response was scored as being either correct or incorrect. Contingency tables were prepared for each of the GCS assessment headings, with the exception of pupil reaction. Two-dimensional tables were produced; response (correct vs incorrect) as row variables and grade (staff nurse, enrolled nurse and student) as column variables. The scores entered into the table cells corresponded to the total number of correct (or incorrect) responses for each grade of staff for all 12 patients assessed. A Chi-square analysis was performed on each table, the null hypotheses being that there were no associations between response and grade. (Chi-square is a statistical formula in which observed response frequencies are compared with frequencies which might be expected by chance, to see if there are significant differences i.e. differences unlikely to occur by chance.)
Correct
201
111
142
Incorrect
123
81
122
V’ (Cramer) = 0.0052; X’ = 4.085; df = 2; p = 0.1297.
These non-significant findings suggest that there is no association between the grade of nurse assessing the patient and the proportion of correct/incorrect responses obtained.
Statistically
significant
findings
The following assessment areas were statistically significant at the 0.05 level. Best motor response
Correct Incorrect
Staff nurses
Enrolled
259* (226)
I20 (134)
nurses
Students 166 (184)
ti5* 198)
72 f58)
93* (80)
NB. Expected values are in parentheses and values with an asterisk indicate a statistically significant departure from independence at the 0.05 level. V’ (Cramer) = 0.034; X’ = 26.68; df = 2; p = 0.00.
INTENSIVE AND CRITICAL
This indicates an association between response (correct/incorrect) and grade of staff. Analysis of residuals using the Freeman-Tukey method identified three cells having values which departed significantly from independence. This suggests that staff nurses made significantly more correct responses and made less errors, while student nurses made more errors than was expected. Right arm Staff nurses Correct Incorrect
174 Cl541 150 (170)
Enrolled nurses
Students
85 190)
110 (1251
105 (100)
154 (138)
NB. Expected values are in parentheses V” (Cramer) = 0.118; X2 = 9.18; df = 2; p = 0.0101.
Right leg Correct Incorrect
Staff nurses
Enrolled nurses
210 (192)
114
126
(114)
(144) 118
114 (132)
students
(77:)
(100)
NB. Expected values are in parentheses V’ (Cramer) = 0.0132; X2 = 10.00; df = 2; p = 0.0067.
For both right arm and leg, no cell was statistically different from the independence model at the 0.05 level using the FreemanHowever, when orthogonal Tukey method. shadow tables were prepared for the right arm (X’ = 8.6847; df = 2; p = 0.013), and right leg (X2 = 9.9289; df = 2; p = 0.007) statistically significant associations were found between the response scores of staff nurses and student nurses. This suggests that there was an association between the proportion of correct/incorrect scores and the grade of staff nurse and student. Staff nurses obtained a higher proportion of- correct responses than expected by chance while students obtained a higher proportion of incorrect responses.
Log-linear analysis It was realised that looking at two-dimensional tables gave specific information on a particular
CARE NURSING
97
assessment, but that it was possible to build a information about the model combining response (correct/incorrect), grade of staff and body part being assessed. A log-linear analysis was made of the data, and the following results were obtained.
Model
L2
df
Saturated [RG] + [RP] + (GP]
0.00 13.45 430.39 38.93 451.60 429.00 16.76 37.98 452.56
10 25 22 17 15 20 12 27
[PI + VI [G] + [PR] [R] + [GP] [RG] + [GP] [RG] + [RP] [RP] + [PG] Independence
0
Significance (%) 100 20 1.4
66.8
Where: R = response, P = part being assessed, (; = grade of staff
The pairwise association model ([KG] + [KP] + [GP]) offers a good representation of the data, however, the [RG] + [RP] model is both more simple and a closer match to the data. (If the model were a ‘perfect’ representation of the data the percentage significance would be 100.) This model indicates that there are associations between response (correctiincorrect), grade of staff and part being assessed. It also suggests there is not an overall association between grade and part being assessed.
GCS assessment errors It was important to consider the nature of the errors made by staff. In the field of neurosurgery some assessment errors can bejudged aS more ‘serious’ than others. A major difficulty was to identify the degree of weakness observed in patients. In the case of 1 patient who had a slight weakness in 1 arm and leg, 52% of KGNs, 50% of ENS and 50% of students marked this as normal power. It was more difficult to establish the pattern of difficulties experienced by staff when assessing severe weakness and flexion. Considering the 2 patients with severe left arm weakness, 40% of’RNs, 44% of ENS and 36% of students marked flexion. In a further 2 patients with flexion, 22% of RGNs, 35% of ENs and 25% of students marked this as severe weakness. For
98
INTENSIVE AND CRITICAL CARE NURSING
the right arm in 2 patients with mild weakness, 35% of RGNs, 25% of ENS and 36% of students marked this as severe weakness, and a further 33% of RGNs, 47% of ENS and 30% of students marked this as flexion. Conversely for patients with a severe weakness or flexion in their right arm response, the responses were marked up one grade by at least 30% of each grade of staff.
DISCUSSION It can generally be said that RGNs made proportionately less errors than either of the other two groups of subjects. It is difficult to understand however, when considering each assessment part independently, why there were statistically significant (albeit weak) associations between grades of staff and right-sided assessments and not left. The patients participating in the video production had similar weakness but varied only in the side of their deficit. When considering assessment parts and grades of staff together, these differences do not appear to be important in the overall model of the data (no [GP] term in the log-linear analysis) However all errors are serious and have potentially important clinical implications. It could be important to miss a mild weakness and chart it as normal power, as this could be the beginning of a patient’s deterioration which could be reversible if identified early. Marking a flexion as an extension is a serious error because it shows lack of assessment skills. There are several explanations of why mistakes can occur: There is a lack of knowledge of how to elicit the best response from a patient and how to interpret the result. There is no clearly agreed standard criterion for each level on the scale which can be universally known and understood. There is a difference in the quality of stimulus applied by different observers. The issue of errors must be considered, however, in the light of this being an investigation of clinical assessment skills away from the bedside. It is unlikely that patient assessment occurs in isolation and without the opportunity
to discuss findings with colleagues. When considering the difficulties experienced by staff in determining the relative strength of a patient’s response, it is artificial to imagine that this can be achieved with a degree of certainty without touching the patients; tactile skills are important in a nursing assessment. It is also possible that’a consensus among assessors would lead to a different nursing judgement. With these issues in mind, however, there remain questions about the pattern of errors identified in this study. Are they due to a lack of assessment skills, interpretation of findings or difficulty in translating findings into a format to score on the GCS? In reviewing the general performance of the three groups of staff participating in the study, RGNs generally had a higher proportion of correct responses, and as can be expected, student nurses faired less favourably. These findings perhaps confirm that specialised training and experience in neurology/surgery do influence the quality of assessment and reporting.
Assumptions
and limitations
The principle assumption underlying this study is that a panel of experts could agree on a ‘correct’ response, and that it was actually correct. There were also certain limitations to this study. First it was not possible to establish the reliability (test re-test) of the GCS with the nursing subjects. It was assumed that the scale was stable over time. Second technical difficulties were encountered with filming, particularly with adequate lighting for the video camera. Finally, although the sample was representative of the unit, care must be taken when generalising these findings to other neurological/surgical units.
RECOMMENDATIONS Various recommendations this study.
can be made
from
Training education These findings suggest that there is a need for improved training in performing assessments
INTENSIVE AND CRlTICAL CARE NURSING
and education in understanding and interpreting assessment findings. A quick run through the assessment at the bedside with a new nurse is certainly insufficient. The video used in this study can, and does, provide a useful teaching aid for both experienced and new staff.
Standard assessment A need exists for an agreed procedure for the assessment of patients. This has been stated in the literature (Sugiura, Muraoka, Chishiki & Baba, 1983) but has not yet happened. Written guide-lines are required that detail how to apply stimulus, the degree of stimulus, and acceptable parameters for results.
Results chart To obtain an accurate .oicture of a patient’s . condition, improvements in the charting of findings are required. It is generally accepted that a ‘C’ to indicate closed eyes due to swelling or ptosis of the eyelid may be written on the chart, but is it necessary always to score leg responses? It can be argued that leg responses are less reliable due to such factors as spinal reflexes, and that more attention could be paid to arm responses and the general position of the patient.
CONCLUSION There is no doubt that the GCS provides an adequate initial assessment of patients with head
99
injuries, particularly in general hospitals and casualty departments. Errors will occur in the assessment of patients and an understanding of how these can occur will be important if the quality of nursing care is to be improved. While the GCS has become an important tool in the assessment of patients, debate continues about the measurement principles underlying its development and use. Furthermore, attempts have been made to improve upon the GCS with such tools as the Glasgow-Liege Scale (Born, 1988). As part of a continuing effort to refine and improve nursing knowledge and practice, there remains the need to examine the cornerstones of what we do.
References Born J D 1988 The Glasgow-Liege scale. Acta Neurochirurgica 9 1: 1- 11 Braakman R, Avezatt C J J, Maas A, Roe1 M, Schouten H.J H 1977 Inter-observer agreement in the assessing of the motor response of the Glasgow ‘coma’ scale. Clinical Neurology Neurosurgery,80 (2): 106-166 Choi S C. Ward I D. Becker D P 1983 Chart for outcome predyction in severe head injury. Journal of Neurosurgery, 59: 294-297 Grant J, Kinney M, Guzzetta C 1996 A method for validating nursing diagnoses. Advances in Nursing Science, 12 (3): 6.5-74 Neatherlin J S, Brillhart B 1988 Glasgow coma scale scores in the patient post-cardio-pulmonary resuscitation. Journal of Neuroscience Nursing, 20 (2): 104-109 Sugiura H, Muraoka K, Chishiki T, Baba M 1983 The Edinburgh-2 coma scale: A new scale for assessing impaired consciousness. Neurosurgery, 12 (4): 4 I l415
hhsti 0
8,9&99
Aspects of neurosurgical assessment using the Glasgow Coma Scale Anne Ellis and Stephen J. Cavanagh
The Glasgow Coma Scale (GCS) has become a cornerstone of the neurological/ surgical assessment of patients used by both nursing and medical staff. Since its development in the 1970s it has been used in a variety of clinical situations to monitor changes in a number of key neurological functions, including level of consciousness, pupil reaction and limb movement. During this time, however, ther have been suggestions that there are problems with some of the measurement principles underlying its use, whch in part has stimulated the development of otht neuro-assessment tools. Irrespective of measurement device, there is always the possibility of error or incorrect assessment. In the field of neurosurgery, as with other high dependency environments, a patient’s condition can change rapidly. Additionally, there is the association of certain assessment responses with nursing and medical interventions. Thus, accuracy in all aspects of assessment and recording is paramount. Despite the growing body of literature surrounding the GCS, little is known about the pattern of errors made by nursing staff using the GCS to assess neurosurgical patients. This study compared the assessment finding of Registered General Nurses (RGNs), Enrolled Nurses and Student Nurses after viewing videotaped neuro-assessments of patients in a high dependency unit. The criterion for judging the accuracy of subject’s assessments was established by a panel of experts. As expected, RGNs had the highest proportion of correct assessments and students the least. Subjects were identified as having difficulty in determining the relative amounts of weakness that a patient exhibited, and in correctly distinguishing between flexion and extension.
INTRODUCTION The practical needs of comatose and neurologitally impaired patients are varied. As a means to Anne Ellis BSc(Hons) RGN RM, Senior Clinical Nurse Manager, Midland Centre for Neurosurgery and Neurology, Stephen J. Cavanagh RGN MS PhD, Division of Health and Nursing Studies Wolverhampton Polytechnic (Requests for offprints to SC) Manuscript accepted 17 February 1992
94
standardise the vocabulary and assessment of patients following head injury, the Glasgow Coma Scale (GCS) was devised by Teasdale & Jennett in the 1970s. The principle underlying this scale is that it should be simple to use by non-medical health providers and more experienced neurological/surgical staff. The scale has been a focus for considerable medical research into its application as a prognostic indicator following head injury (Choi, Ward SCBecker, 1983), and the inter-rater reliability of staff
INTENSIVEANDCRlTlCALCARENURSlNG
using it (Braakman, Avezatt, Maas, Roe1 8c Schouten, 1977). There remains, however, relatively little nursing research related to the GCS. Yet the studies which have been performed address important nursing care issues, for example, Grant, Kinney & Guzzetta (1990) proposed that the GCS could be used as a means of validating nursing diagnoses, and Neatherlin & Brillhart (1988) used the GCS as an assessment tool with patients following cardio-pulmonary resuscitation. It is recognised that it is not possible to measure directly a concept such as level of consciousness with a single assessment tool. The underlying premise of the GCS is that continuous monitoring under ‘standard conditions’ will lead to a determination of a trend in a patient’s condition which can be more readily interpreted. The majority of this recording is currently undertaken on 24 h basis by either Kegistered General Nurses (KGNs), Enrolled Nurses (ENS) or Student Nurses (SNs) with appropriate supervision. While there has been some standardisation in the assessment and recording of assessment data using the GCS, a neurological/surgical patient can present severe challenges to the clinical judgement of the most experienced practitioner. Misinterpretation of clinical signs is possible as are errors in recording. Many ‘difhcult’cases will be resolved by either a consensus approach or by seeking additional assistance, nevertheless assessment errors will occur. In view of the knowledge that the clinical condition of neurological/ surgical patients can exchange extremely rapidly and that clearly errors can be made, it is perhaps surprising that little is known about the nature of assessment errors made by nursing staff using the GCS. This information will be useful to educators and clinicians alike in identifying improvements in the GCS, the teaching of assessment skills, and adopting the scale in nursing practice.
Purpose The purpose of this study was to investigate the pattern of errors made by RGNs, ENS and SNs when assessing neurosurgical patients.
95
METHODS Sample A random sample of 27 registered and 16 enrolled nurses employed in a regional neurological/neurosurgical centre were used as subjects in the study. In addition, a convenience sample of 22 third-year-student nurses with no previous neuroscience experience were chosen as comparison subjects. 77% of RGNs and 81% of ENS had more than 1 full-time year of neurosurgical experience. In addition, 32% of the RGNs possessed the English National Board (ENB) 148 and 10% the ENB 100 certificates. 12% of ENS had successfully completed the ENB 153 course.
Procedure Before commencing this study permission was obtained from the Ethical Committee, and a letter of consent was completed by each patient or relative involved. A video was made of the assessment of 19 patients, of these 12 were chosen as being representative of a variety of neurological responses, i.e. alert and orientated, no limb deficits, confused, mild limb deficits, severe deficits, flexing or extending to pain, or no response at all. An attempt was made to have patients in the same position (on their backs in bed) and for the nurses carrying out assessments to perform them in the same order. Stimuli were applied in an identical manner (nailbed pressure) to obtain the maximum response. This was not always possible as some patients were not able to remain in a recumbent position. There were also technical difficulties with filming, in particular with the recording of pupil responses; this item was excluded from the study. Of the 12 patients recruited to this study, 2 had no deficits, 5 had left-sided v-roakness, 4 right-sided weakness, and 1 was unresponsive to any stimulation. A panel of experts consisting of 3 registered nurses and 1 senior registrar (all with considerable experience in neurosciences) was assembled to view the video and determine an appropriate assessment response to be recorded on the GCS.
96
INTENSlVEANDCRlTlCALCARENURSlNG
Their agreed ‘standard’ was used as the criterion to rate the responses of the nursing subjects when they viewed the video.
Non-statistically
significant
findings
The following assessment areas produced significant findings at the the 0.5 level.
non-
Prior to the study all subjects were given: 1. an introduction and explanation of the study 2. a demographic questionnaire, including years in neuroscience and any relevant courses 3. a diagram and written explanation of terms (e.g. localise to pain, flexion, extension) 4. a GCS chart for recording results observed on each of the 12 patients. Subjects observed the video in groups of approximately 6. They were given time to read the instructions and asked not to confer. The video was stopped between each patient assessment
to allow
responses.
the
participants
to record
Eye opening
Correct
Enrolled
256
142
220
68
49
46
Incorrect
nurses
Students
V’ (Cramer) = 0.006; X’ = 4.71; df = 2; p = 0.094 Verbal
Correct
Staff nurses
Enrolled
283
155
214
41
37
98
Incorrect
nurses
Students
V’ (Cramer) = 0.007; X2 = 5.70; df = 2; p = 0.057 Left arm
their
It took 45 min to assess all patients.
Staff nurses
Staff nurses Correct Incorrect V2 (Cramer)
171 153 = 0.0054;
Enrolled nurses
Students
86
121
106
143
X’ = 4.18; df = 2; p = 0.1236.
RESULTS AND INTERPRETATION
Left leg Staff nurses
Enrolled nurses
Students
A tally was made of subjects’ assessments based on the video, and these were judged against the criteria agreed by the panel of experts. For the purposes of analysis, an individual’s response was scored as being either correct or incorrect. Contingency tables were prepared for each of the GCS assessment headings, with the exception of pupil reaction. Two-dimensional tables were produced; response (correct vs incorrect) as row variables and grade (staff nurse, enrolled nurse and student) as column variables. The scores entered into the table cells corresponded to the total number of correct (or incorrect) responses for each grade of staff for all 12 patients assessed. A Chi-square analysis was performed on each table, the null hypotheses being that there were no associations between response and grade. (Chi-square is a statistical formula in which observed response frequencies are compared with frequencies which might be expected by chance, to see if there are significant differences i.e. differences unlikely to occur by chance.)
Correct
201
111
142
Incorrect
123
81
122
V’ (Cramer) = 0.0052; X’ = 4.085; df = 2; p = 0.1297.
These non-significant findings suggest that there is no association between the grade of nurse assessing the patient and the proportion of correct/incorrect responses obtained.
Statistically
significant
findings
The following assessment areas were statistically significant at the 0.05 level. Best motor response
Correct Incorrect
Staff nurses
Enrolled
259* (226)
I20 (134)
nurses
Students 166 (184)
ti5* 198)
72 f58)
93* (80)
NB. Expected values are in parentheses and values with an asterisk indicate a statistically significant departure from independence at the 0.05 level. V’ (Cramer) = 0.034; X’ = 26.68; df = 2; p = 0.00.
INTENSIVE AND CRITICAL
This indicates an association between response (correct/incorrect) and grade of staff. Analysis of residuals using the Freeman-Tukey method identified three cells having values which departed significantly from independence. This suggests that staff nurses made significantly more correct responses and made less errors, while student nurses made more errors than was expected. Right arm Staff nurses Correct Incorrect
174 Cl541 150 (170)
Enrolled nurses
Students
85 190)
110 (1251
105 (100)
154 (138)
NB. Expected values are in parentheses V” (Cramer) = 0.118; X2 = 9.18; df = 2; p = 0.0101.
Right leg Correct Incorrect
Staff nurses
Enrolled nurses
210 (192)
114
126
(114)
(144) 118
114 (132)
students
(77:)
(100)
NB. Expected values are in parentheses V’ (Cramer) = 0.0132; X2 = 10.00; df = 2; p = 0.0067.
For both right arm and leg, no cell was statistically different from the independence model at the 0.05 level using the FreemanHowever, when orthogonal Tukey method. shadow tables were prepared for the right arm (X’ = 8.6847; df = 2; p = 0.013), and right leg (X2 = 9.9289; df = 2; p = 0.007) statistically significant associations were found between the response scores of staff nurses and student nurses. This suggests that there was an association between the proportion of correct/incorrect scores and the grade of staff nurse and student. Staff nurses obtained a higher proportion of- correct responses than expected by chance while students obtained a higher proportion of incorrect responses.
Log-linear analysis It was realised that looking at two-dimensional tables gave specific information on a particular
CARE NURSING
97
assessment, but that it was possible to build a information about the model combining response (correct/incorrect), grade of staff and body part being assessed. A log-linear analysis was made of the data, and the following results were obtained.
Model
L2
df
Saturated [RG] + [RP] + (GP]
0.00 13.45 430.39 38.93 451.60 429.00 16.76 37.98 452.56
10 25 22 17 15 20 12 27
[PI + VI [G] + [PR] [R] + [GP] [RG] + [GP] [RG] + [RP] [RP] + [PG] Independence
0
Significance (%) 100 20 1.4
66.8
Where: R = response, P = part being assessed, (; = grade of staff
The pairwise association model ([KG] + [KP] + [GP]) offers a good representation of the data, however, the [RG] + [RP] model is both more simple and a closer match to the data. (If the model were a ‘perfect’ representation of the data the percentage significance would be 100.) This model indicates that there are associations between response (correctiincorrect), grade of staff and part being assessed. It also suggests there is not an overall association between grade and part being assessed.
GCS assessment errors It was important to consider the nature of the errors made by staff. In the field of neurosurgery some assessment errors can bejudged aS more ‘serious’ than others. A major difficulty was to identify the degree of weakness observed in patients. In the case of 1 patient who had a slight weakness in 1 arm and leg, 52% of KGNs, 50% of ENS and 50% of students marked this as normal power. It was more difficult to establish the pattern of difficulties experienced by staff when assessing severe weakness and flexion. Considering the 2 patients with severe left arm weakness, 40% of’RNs, 44% of ENS and 36% of students marked flexion. In a further 2 patients with flexion, 22% of RGNs, 35% of ENs and 25% of students marked this as severe weakness. For
98
INTENSIVE AND CRITICAL CARE NURSING
the right arm in 2 patients with mild weakness, 35% of RGNs, 25% of ENS and 36% of students marked this as severe weakness, and a further 33% of RGNs, 47% of ENS and 30% of students marked this as flexion. Conversely for patients with a severe weakness or flexion in their right arm response, the responses were marked up one grade by at least 30% of each grade of staff.
DISCUSSION It can generally be said that RGNs made proportionately less errors than either of the other two groups of subjects. It is difficult to understand however, when considering each assessment part independently, why there were statistically significant (albeit weak) associations between grades of staff and right-sided assessments and not left. The patients participating in the video production had similar weakness but varied only in the side of their deficit. When considering assessment parts and grades of staff together, these differences do not appear to be important in the overall model of the data (no [GP] term in the log-linear analysis) However all errors are serious and have potentially important clinical implications. It could be important to miss a mild weakness and chart it as normal power, as this could be the beginning of a patient’s deterioration which could be reversible if identified early. Marking a flexion as an extension is a serious error because it shows lack of assessment skills. There are several explanations of why mistakes can occur: There is a lack of knowledge of how to elicit the best response from a patient and how to interpret the result. There is no clearly agreed standard criterion for each level on the scale which can be universally known and understood. There is a difference in the quality of stimulus applied by different observers. The issue of errors must be considered, however, in the light of this being an investigation of clinical assessment skills away from the bedside. It is unlikely that patient assessment occurs in isolation and without the opportunity
to discuss findings with colleagues. When considering the difficulties experienced by staff in determining the relative strength of a patient’s response, it is artificial to imagine that this can be achieved with a degree of certainty without touching the patients; tactile skills are important in a nursing assessment. It is also possible that’a consensus among assessors would lead to a different nursing judgement. With these issues in mind, however, there remain questions about the pattern of errors identified in this study. Are they due to a lack of assessment skills, interpretation of findings or difficulty in translating findings into a format to score on the GCS? In reviewing the general performance of the three groups of staff participating in the study, RGNs generally had a higher proportion of correct responses, and as can be expected, student nurses faired less favourably. These findings perhaps confirm that specialised training and experience in neurology/surgery do influence the quality of assessment and reporting.
Assumptions
and limitations
The principle assumption underlying this study is that a panel of experts could agree on a ‘correct’ response, and that it was actually correct. There were also certain limitations to this study. First it was not possible to establish the reliability (test re-test) of the GCS with the nursing subjects. It was assumed that the scale was stable over time. Second technical difficulties were encountered with filming, particularly with adequate lighting for the video camera. Finally, although the sample was representative of the unit, care must be taken when generalising these findings to other neurological/surgical units.
RECOMMENDATIONS Various recommendations this study.
can be made
from
Training education These findings suggest that there is a need for improved training in performing assessments
INTENSIVE AND CRlTICAL CARE NURSING
and education in understanding and interpreting assessment findings. A quick run through the assessment at the bedside with a new nurse is certainly insufficient. The video used in this study can, and does, provide a useful teaching aid for both experienced and new staff.
Standard assessment A need exists for an agreed procedure for the assessment of patients. This has been stated in the literature (Sugiura, Muraoka, Chishiki & Baba, 1983) but has not yet happened. Written guide-lines are required that detail how to apply stimulus, the degree of stimulus, and acceptable parameters for results.
Results chart To obtain an accurate .oicture of a patient’s . condition, improvements in the charting of findings are required. It is generally accepted that a ‘C’ to indicate closed eyes due to swelling or ptosis of the eyelid may be written on the chart, but is it necessary always to score leg responses? It can be argued that leg responses are less reliable due to such factors as spinal reflexes, and that more attention could be paid to arm responses and the general position of the patient.
CONCLUSION There is no doubt that the GCS provides an adequate initial assessment of patients with head
99
injuries, particularly in general hospitals and casualty departments. Errors will occur in the assessment of patients and an understanding of how these can occur will be important if the quality of nursing care is to be improved. While the GCS has become an important tool in the assessment of patients, debate continues about the measurement principles underlying its development and use. Furthermore, attempts have been made to improve upon the GCS with such tools as the Glasgow-Liege Scale (Born, 1988). As part of a continuing effort to refine and improve nursing knowledge and practice, there remains the need to examine the cornerstones of what we do.
References Born J D 1988 The Glasgow-Liege scale. Acta Neurochirurgica 9 1: 1- 11 Braakman R, Avezatt C J J, Maas A, Roe1 M, Schouten H.J H 1977 Inter-observer agreement in the assessing of the motor response of the Glasgow ‘coma’ scale. Clinical Neurology Neurosurgery,80 (2): 106-166 Choi S C. Ward I D. Becker D P 1983 Chart for outcome predyction in severe head injury. Journal of Neurosurgery, 59: 294-297 Grant J, Kinney M, Guzzetta C 1996 A method for validating nursing diagnoses. Advances in Nursing Science, 12 (3): 6.5-74 Neatherlin J S, Brillhart B 1988 Glasgow coma scale scores in the patient post-cardio-pulmonary resuscitation. Journal of Neuroscience Nursing, 20 (2): 104-109 Sugiura H, Muraoka K, Chishiki T, Baba M 1983 The Edinburgh-2 coma scale: A new scale for assessing impaired consciousness. Neurosurgery, 12 (4): 4 I l415