Impact of systematic evaluation of pain and agitationin an intensive care unit*

Gerald Chanques, MD; Samir Jaber, MD, PhD; Eric Barbotte, MD; Sophie Violet, RN;Mustapha Sebbane, MD; Pierre-François Perrigault, MD; Claude Mann, MD, PhD;Jean-Yves Lefrant, MD, PhD; Jean-Jacques Eledjam, MD, PhD

Medical societies have rec-ommended the evaluationof pain and agitation levelsand the titration of sedative

and analgesics drugs in intensive careunit (ICU) patients (1, 2) consistent withthe Joint Commission on Accreditation ofHealthcare Organization’s standards (3,4). Based on the European survey (5), ofthe 43% of ICUs that do use sedationscales, 74% use the Ramsay scale al-though this scale does not evaluate agi-

tation states (6, 7). Management of painseems to also be insufficient. Painful pro-cedures were performed after administra-tion of an analgesic in �20–40% (8, 9).ICU physicians are certainly uncomfort-able treating pain appropriately (10) be-cause of organ system dysfunction, im-paired mental status, and alteredpharmacokinetics and pharmacodynam-ics apropos of the critically ill population.Specifically, this population is more sus-ceptible to the side effects of analgesics orsedative drugs (vasoplegia, respiratorydepression, inhibition of cough, constipa-tion, vigilance impairment). Analgesicscan be associated with adverse events. Ithas been reported that they can be asso-ciated with an increased length of venti-lation and stay in the ICU (11).

Surprisingly, there are no publishedstudies about the impact of a systematicevaluation of pain and agitation in ICU, asrecommended in official practice guide-lines (1–3, 7).

We conducted the present before-afterstudy to test the hypothesis that imple-mentation of a systematic evaluation ofpain and agitation by nurse following by amedical intervention in an ICU would beassociated with a decrease in the inci-dence and intensity of pain and agitation.

MATERIALS AND METHODS

Population

The present before-after study took placein the 12-bed medical-surgical ICU of the StEloi Hospital, a 660-bed teaching and referralfacility of the University of Montpellier inFrance, staffed by 35 nurses, 20 assistantnurses, five physicians, and three residents. Allconsecutive patients �18 yrs old and stayingin the ICU for �24 hrs were eligible. Only thefirst admission in the ICU was included duringeach phase. Exclusion criteria were decision towithdraw life support within 48 hrs after ad-mission, brain injuries that limited communi-cation by the patient, transfer to another ICU

*See also p. 1838.From the Intensive Care and Anesthesiology De-

partment “B” (DAR B), Saint Eloi Hospital, MontpellierUniversity Hospital, Montpellier Cedex 5, France (GC,SJ, SV, MS, P-FP, CM, J-YL, J-JE); and Department ofMedical Statistics, Arnaud de Villeneuve Hospital,Montpellier University Hospital, Montpellier Cedex 5,France (EB).

The authors do not have any financial interests todisclose.

Copyright © 2006 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/01.CCM.0000218416.62457.56

Objective: To measure the impact of implementation of thesystematic evaluation of pain and agitation by nurses using theBehavioral Pain Scale (BPS), the Numerical Rating Scale (NRS) forpain, and the Richmond Agitation Sedation Scale (RASS) associ-ated with medical staff education in analgesia and sedationmanagement in intensive care unit (ICU) patients.

Design: Two-phase, prospective, controlled study.Setting: Twelve-bed medical-surgical ICU in a university hos-

pital.Patients: Consecutive patients staying >24 hrs in ICU.Interventions: BPS, NRS, and RASS were measured twice daily,

at rest, by independent observers during 21 wks (control group)and after 4 wks of training, by nurses during 29 wks (interventiongroup). In the intervention group, the treating physician wasalerted in case of pain defined by BPS >5 or NRS >3 or in caseof agitation defined by RASS >1.

Measurements and Main Results: A total of 230 patients wereincluded (control group, n � 100; intervention group, n � 130).Baseline characteristics were not significantly different. The in-

cidence of pain and agitation decreased significantly in theintervention group: 63% vs. 42% (p � .002) and 29% vs. 12%(p � .002), respectively. Rate of severe pain and agitation eventsdefined by NRS >6 and RASS >2, respectively, also decreasedsignificantly. There were significantly more therapeutic changesin the intervention group in the way of an escalation but also inthe way of a de-escalation for analgesic and psychoactive drugs.Compared with the control group, there was a marked decrease inthe duration of mechanical ventilation (120 [interquartile range48–312] vs. 65 (24–192) hrs, p � .01) and nosocomial infectionsrate (17% vs. 8%, p < .05) in the intervention group. There wasno significant difference in median length of stay (9 [4, 15] vs. 7[4, 13] days) and mortality in ICU (12 vs. 15%).

Conclusions: Systematic evaluation of pain and agitation, andanalgesics and sedatives need was associated with a decrease inincidence of pain and agitation, duration of mechanical ventilationand nosocomial infections. (Crit Care Med 2006; 34:1691–1699)

KEY WORDS: pain; analgesia; agitation; sedation; intensive careunit; practice guidelines

1691Crit Care Med 2006 Vol. 34, No. 6

for specialized care, and stay in the ICU duringboth phases of the study.

Ethics and Consent

The Ethics Committee of the French Soci-ety of Critical Care waived the need for in-formed consent.

Study Design

Control Phase. During the control phase(from November 2002 to March 2003, 21 wks)no systematic and objective evaluation of painor agitation was done by nurses or physicians.Six independent observers (students in medi-cine and pharmacy) evaluated pain and agita-tion levels among all the patients admitted inthe unit (tools described subsequently). Thisevaluation occurred at rest, 30 mins after anyprocedure, between the times of 0800 and1000 hrs, and between 2000 and 2200 hrs.Nurses and treating physicians were blinded tothe results of pain and agitation evaluations.The same instructor (GC) performed a stan-dardized individual training for observers atthe bedside on ten patients. Then they weretested for reliability in 43 patients during 10evaluation days (October 2002). Intraclass co-efficient correlation was .97 (upper 95th per-cent confidence limit, .99) for the RichmondAgitation Sedation Scale (RASS) and .81 (.93)for the Behavioral Pain Scale (BPS).

Interphase. Physicians and residents re-ceived an oral education and a written support(Appendix 1) that encouraged them to vigor-ously assess and treat pain and/or agitationafter recognition of a significant event bynurses using a systematic approach: confirmthe presence of pain/agitation, diagnose thesource, choose the appropriate analgesic/psychoactive drug, and consider the risk-benefit ratio of drug administration accordingto each clinical situation. Treatment was or-dered only by the treating physician. On dailyrounds (between the times of 1000 and 1300hrs and between 2200 and 0100 hrs), physi-cians considered the necessity to maintain an-algesic and/or psychoactive treatments andwere encouraged to decrease doses to the min-imal effective dose (Appendix 1). During 4 wks(March 2003) the 35 nurses were trained in-dividually at the bedside by the same instruc-tor to evaluate pain and agitation levels in thesame way as that done by the control phaseobservers.

Intervention Phase. During the interven-tion phase (from April to October 2003, 29wks), bedside nurses evaluated pain and agita-tion levels among all the patients admitted tothe ICU. This evaluation occurred at rest, 30mins after any procedure, between the timesof 0800 and 1000 hrs, between 1400 and 1600hrs, and between 2000 and 2200 hrs, such thateach nursing shift had an identified 2-hr pe-riod during which assessments were done.Nurses could perform additional evaluations

at any time of the day or night and duringpainful procedures. The physician waspromptly notified of a pain or an agitationevent, regardless of the hour. Informationabout these evaluations was communicatedduring each daily meeting between nurses andphysicians. Nurses began the clinical presen-tation of their patient with evaluation of painand agitation levels, which could be consid-ered as a “vital sign” (4) along with usual vitalsigns (heart rate, blood pressure, respiratoryrate, oximetry, urine output, and body tem-perature). Education performed during the in-terphase was continued during daily meetingbetween nurses and physicians.

Evaluation of Pain and Agitation. Choiceof pain and agitation evaluation tools was per-formed after review of the literature (2, 12, 13)and pilot tests. The pain evaluation tool usedwas the 0–10 numerical relative scale (NRS,appendix 2) (12). This scale was adapted to ICUpatients, who often suffer from sensorial defi-ciencies; therefore, the printed scale was en-larged to be easily visible (3.9 � 11.8 inches).The BPS (Appendix 2) (14) was used for eval-uation of pain in intubated patients if theywere not able to perform the NRS.

Two hundred and twenty evaluations wereperformed independently during a period �1hr on 38 patients to measure the reliabilitybetween independent observers and nurses.The RASS (Appendix 2) (15, 16) was used toevaluate the agitation level. A French trans-lated version, which had been validated previ-ously (17), was used. Intraclass coefficient cor-relation was .96 (upper 95th percent confidencelimit, .97) for the RASS, .86 (.91) for the BPS,and .90 (.94) for the NRS. Reliability of NRSwas tested because patients could worry aboutpain differently with nurses than with inde-pendent observers.

A pocket-card was given to control phaseobservers and intervention phase nurses aftertraining. Laminated plastic posters with thelarge-print ICU-adapted NRS were placed onthe wall of each room. All of these tools wereapproved by the infection control departmentof Montpellier University Hospital accordingto usual guidelines (18) and created by thecommunications department of the hospital.No financial support was provided by pharma-ceutical industries.

A pain event was defined by either a BPSscore �5 (14) or a NRS level �3 (12). Anagitation event was defined by a RASS level �1(15, 16).

Weaning Practices. Weaning practices didnot differ between the two phases for patientswho required mechanical ventilation. Patientswere considered candidates for weaning whenthey no longer had high-grade fever, hemody-namic instability, or severely altered con-sciousness, as well as exhibiting adequate ox-ygenation with an FIO2 50% and positive end-expiratory pressure �5 cm H2O.

Candidates for weaning were switched topressure support ventilation and underwentdaily spontaneous breathing trials on a T piece.

Decision to extubate was based on simple bed-side tolerance variables, including respiratoryrate, SpO2, and the use of accessory respira-tory muscles during T-piece trials.

Data Collection

Data were prospectively recorded duringboth phases of the study. Age, gender, type ofadmission, and the Simplified Acute Physiol-ogy Score II (19) were collected within 24 hrsafter admission. Medical admission was de-fined by the absence of surgical intervention.Planned surgery was defined by admission tothe ICU immediately after a surgical interven-tion that was planned �24 hrs in advance.Unplanned surgery was defined by admissionto the ICU immediately after an unplannedintervention. Postoperative complication wasdefined by an admission from surgical ward tothe ICU without prior reintervention. Numberand level of events of pain and agitation wererecorded daily during the two periods of eval-uation common to both phases: at rest, be-tween the times of 0800 and 1000 hrs, andbetween 2000 and 2200 hrs. The median ob-servation rate of the systematic evaluation ofpain and agitation levels during both phaseswas defined as follows: median of (the numberof evaluations done between the times of 0800and 1000 hrs and between 2000 and 2200 hrs �100/the number of evaluations which shouldhave been done, i.e., the length of stay in theICU � 2). Severe pain events were defined byeither a BPS score �7 or an NRS level �6(12). Severe agitation events were defined by aRASS level �2 (15). Throughout the period ofstay in the ICU, we recorded the day and timeof initiation of mechanical ventilation; day andtime of end of mechanical ventilation, whichwas considered the day the patient was extu-bated, provided that there was no need forreintubation within the next 48 hrs; day andtime of initiation of intravenous continuoussedation; type of psychoactive drugs and anal-gesics classified according the three levels ofthe World Health Organization (WHO); dailydose of continuous intravenous sedation (mi-dazolam or propofol � fentanyl or morphine);total dose of other intravenous psychoactivedrugs and analgesics; number of therapeuticchanges in the way of an escalation defined byan administration of a new drug and/or anincrease of �20% of the daily dose; and num-ber of therapeutic changes in the way of ade-escalation defined by an interruption of adrug and/or a decrease of �20% of the dailydose. Throughout the stay in ICU, we recordedthe occurrence of adverse events: surgical re-intervention, myocardial ischemia, thrombo-embolic event, intestinal occlusion requiringsurgery or colo-aspiration, self-extubation,and self-removal of central venous catheter,bladder catheter, or gastric tube. At time of

1692 Crit Care Med 2006 Vol. 34, No. 6

discharge we collected the first day of intesti-nal transit, the maximal amount of daily gas-tric residuals, and data related to nosocomialinfections expressed as the ratio of nosocomialinfections to the number days exposed to therisk and the number of patients having a nos-ocomial infection (ventilator-acquired pneu-monia: at least one organism isolated by bron-choalveolar lavage at a concentration �104

colony-forming units (CFU)/mL; colonization ofcentral venous catheters: at least one organ-ism at a concentration �103 CFU/mL identi-fied by a culture of the catheter tip via theBrun-Buisson technique (20); urinary cathe-ter related infection: the association of a leu-kocyturia at a concentration of �104 mL withthe presence of an organism at a concentra-tion of 105 CFU/mL; bacteremia: a positivehemoculture with the isolation of an organ-ism or at least two positive hemocultures for acoagulase-negative Staphylococcus accordingto the usual definitions (21). Length of stayand mortality in the ICU were recorded.

Questionnaire

A questionnaire was administered anony-mously to nurses and assistant nurses beforeand after intervention phase. They were askedto quantify the quality of pain and agitationmanagement in the ICU (inadequate, moder-ate, well, or very well) and to specify the con-tributing factors that should be improved.

End Points

The primary end point was the incidence ofpain defined by the proportion of patients whodeveloped at least one pain event (BPS �5and/or NRS �3) and the incidence of agitationdefined by the proportion of patients who de-veloped at least one agitation event (RASS �1)during their ICU stay, at rest, between thetimes of 0800 and 1000 hrs and between 2000and 2200 hrs. Secondary end points were a)the duration of mechanical ventilation, lengthof stay, adverse outcomes, and mortality inICU; b) changes in analgesics and psychoactiveprescriptions practices; and c) nurse satisfac-tion with pain and agitation management inthe ICU.

Statistical Analysis

With an incidence of pain of 60% and ag-itation of 30%, the sample size needed to showa 30% reduction of incidence of pain or agita-tion, with � and � errors of .05 and .20,respectively, would be 99 analyzable patientsin each study period. Considering ineligiblepatients, this translated to 120 admitted pa-tients in each phase of 5 months.

Quantitative data were shown as mediansand 25th–75th percentiles. Mann-Whitney U test

(quantitative data) and Fisher’s test (qualita-tive data) were used to compare patients in-cluded in the control phase (control group) tothose in the intervention phase (interventiongroup). Relative risks to not develop a pain oran agitation event, associated with interven-tion, were calculated and expressed with their95% confidence limits. We considered p � .05to be statistically significant. Data were ana-lyzed using the SAS software version 6.12(SAS Institute, Cary, NC). The Department ofMedical Statistics of the Montpellier Univer-sity Hospital performed statistical analysis.

RESULTS

Two hundred and ninety five patientswere admitted during the period of thestudy. During the control phase, 122 pa-tients were admitted a total of 125 timesin the ICU. Seventeen patients were notincluded because they stayed �24 hrs inthe ICU. Five patients were excluded be-cause of a decision to withdraw life sup-port within 48 hrs after admission (n �2), impossible communication (n � 1),transfer to another ICU (n � 1), or stay in

Table 1. Patients’ characteristics

Control Group(n � 100)

Intervention Group(n � 130) p Value

Age, yrs 58 (51–74) 59 (48–71) .54Female gender, n (%) 36 (36) 41 (32) .48Weight, kg 70 (60–76) 67 (60–84) .86SAPS II 32 (24, 41) 31 (21, 41) .53Admission type .56

Unplanned surgery, n (%) 25 (25) 27 (20)Planned surgery, n (%) 18 (18) 29 (23)Liver transplantation, n (%) 10 (10) 15 (11)Postoperative complication, n (%) 11 (11) 7 (6)Medical, n (%) 32 (32) 44 (34)Trauma, n (%) 4 (4) 8 (6)

Mechanical ventilation at admission, n (%) 61 (61) 86 (66) .42Mechanical ventilation during ICU stay, n (%) 66 (66) 94 (72) .78

SAPS, Simplified Acute Physiology Score (19); ICU, intensive care unit.Quantitative data were expressed as median (25th–75th percentiles).

Table 2. Evaluation of pain and agitation

Control Group(n � 100)

Intervention Group(n � 130) p Value

Evaluation of pain (BPS and NRS)Total number of ratings during total

ICU stay1,442 2,016

No. of ratings per patient during totalICU stay

8.0 (4.0–15.3) 11.0 (5.0–18.5) .10

No. of ratings per patient per day of ICU stay 1.2 (0.8–1.5) 1.5 (1.2–1.8) �.001Observation rate of the systematic

evaluation of pain, %58.6 (40.0–76.7) 75.0 (62.0–90.0) �.001

Evaluation of agitation (RASS)Total number of ratings during total

ICU stay1,627 2,124

Number of ratings per patient during totalICU stay

10.0 (4.0–19.0) 12.0 (6.0–19.0) .25

Number of ratings per patient per day ofICU stay

1.3 (1.0–1.6) 1.6 (1.4–1.8) �.001

Observation rate of the systematicevaluation of agitation (%)

64.0 (50.0–81.7) 77.1 (70.0–90.8) �.001

BPS, Behavioral Pain Scale; NRS, Numerical Rating Scale; ICU, intensive care unit; RASS,Richmond Agitation Sedation Scale.

The median observation rate of the systematic evaluation of pain and agitation levels during bothphases was defined as follows: median of (the number of evaluations done between the times of 0800and 1000 hrs and between 2000 and 2200 hrs � 100/the number of evaluations that should have beendone, i.e. the length of stay in the ICU � 2). No significant difference was observed in the rate of BPSratings among pain ratings (BPS and NRS) performed in ventilated patients during their ICU stay (39.4[6.0–71.0] vs. 36.1 [11.3–65.7]%, p � .98).

Quantitative data were expressed in median (25th–75th percentiles).

1693Crit Care Med 2006 Vol. 34, No. 6

the ICU during both phases (n � 1). Dur-ing the intervention phase, 173 patientswere admitted a total of 180 times in theICU. Thirty-four patients were not in-cluded because they stayed �24 hrs inthe ICU. Six patients were excluded be-cause of a decision to withdraw life sup-port within 48 hrs after admission (n � 5)or impossible communication (n � 1).Three patients were not included becauseof missing data. Thus, a total of 230 pa-tients were included for analysis: 100 pa-tients admitted 100 times (control group)and 130 patients admitted 130 times (in-tervention group).

Patients characteristics of the twogroups were similar (Table 1). The me-dian observation rate of systematic eval-uation of pain and agitation was signifi-cantly higher in the intervention groupthan in the control group (Table 2).

However, compared with the controlgroup, the incidence of pain and agitationwas significantly lower in the interven-tion group: 63 vs. 42% (p � .002) and 29vs. 12% (p � .002), respectively (Fig. 1).In the intervention group, the relativerisk to not develop a pain event was 1.44(1.14, 1.81) and the relative risk to notdevelop an agitation event was 1.73 (1.15,2.61). The incidence of severe pain de-fined by an NRS level �6 or a BPS level�7 and severe agitation defined by aRASS level �2 were significantly lower inthe intervention group: 36 vs. 16% (p � .001)and 18 vs. 5% (p � .002), Rate of painevaluated by BPS among patients whodeveloped a pain event was not signifi-cantly different between the two groups:14 of 63 (22%) vs. ten of 55 (18%), p � .65,respectively.

Among pain and agitation events, therate of severe pain events evaluated by anNRS level �6 and the rate of severe agi-tation events evaluated by a RASS level�2 were significantly lower in the inter-vention group: 57 of 176 vs. 24 of 168events (p � .001) and 42 of 82 vs. 9 of 31events (p � .03) respectively (Fig. 2). Therate of severe pain events evaluated by aBPS level �7 was not significantly differ-ent between the two groups: 12 of 38 vs.four of 17 (p � .75, Fig. 2).

Main adverse events are shown in Ta-ble 3 and nosocomial infections in Table4. Compared with the control group,there was a marked decrease in durationof mechanical ventilation and nosocomialinfections rate in the intervention group(Tables 3 and 4). There was no significantdifference in median length of stay and

mortality in ICU between the two groups(Table 3).

Among the analgesic and sedativedrugs used during the two phases, tram-adol was the only drug used significantlymore frequently in the interventiongroup (16 vs. 27%, p � .05). There was nosignificant difference among drugs usedfor continuous sedation in the twogroups. Midazolam, propofol, or eitherwas used for continuous sedation respec-tively in 87%, 7%, and 6% in controlgroup vs. 74%, 16%, and 10% in theintervention group (p � .18). Fentanyl,morphine, or either was used for contin-

uous sedation respectively in 89%, 2%,and 9% in control group vs. 80%, 8%,and 12% in the intervention group (p � .31).The dose of morphine administered ascontinuous sedation showed a trend tobe higher in the intervention group: 0.1(0.1–0.5) vs. 0.3 (0.4–1.5) mg/kg per dayof sedation, p � .06. The doses of clonidineand the spasmolytic intestinal drug phlo-roglucinol were significantly higher inthe intervention group: 12 (6–28) vs. 31(15–66) �g per day of stay in the ICU(p � .01) and 31 (9–67) vs. 147 (68–225)mg per day of stay in the ICU (p � .01),respectively.

Figure 1. Incidence of pain and agitation in the two groups. This figure shows the decrease of the incidenceof pain and agitation between group 1 (n � 100) and group 2 (n � 130). **p � .01.

Figure 2. Rate of severe pain and agitation events among painful and agitated patients. This figureshows the decrease between the two groups in the rate of severe pain events (defined by a BehavioralPain Scale [BPS] �7 or a Numerical Rating Scale [NRS] �6) and severe agitation events (defined bya Richmond Agitation Sedation Scale [RASS] �2) among significant pain events defined by a BPS �5(control group, 38 events; intervention group, 17) or a NRS �3 (control group, 176 events; interven-tion group, 168) and significant agitation events defined by a RASS �1 (control group, 82 events;intervention group, 31). NS, not significantly different. *p � .05. ***p � .001.

1694 Crit Care Med 2006 Vol. 34, No. 6

There were more therapeutic changesin the intervention group in the way of anescalation but also in the way of a de-escalation for third WHO level analgesics,centrally acting muscle relaxants, spas-molytic intestinal and nonsteroid anti-inflammatory drugs, neuroleptics, andclonidine (Table 5).

Twenty-five nurses answered the ques-tionnaire before the intervention phaseand 32 after. Rate of satisfaction and con-tributing factors of dissatisfaction withpain and agitation management areshown in Figures 3 and 4.

DISCUSSION

The main finding of our study is thatsystematic evaluation of pain and agita-tion levels by nurses with rapid call to aphysician in case of pain or agitation de-creased the observed incidence and inten-sity of pain and agitation in ICU patients.This improvement in pain and agitationmanagement was associated with a betteroutcome (shorter duration of mechanicalventilation and lower nosocomial infec-tions rate). These results could be ex-plained by a better match between anal-gesic and psychoactive drugs and pa-tients’ needs for these drugs. Educationof physicians and daily meeting betweennurses and physicians reinforced the im-portance of rapid physician response tonurses’ calls and vigorous evaluation andtreatment of the reported pain and agita-tion. Medical intervention delay, themain cause of nurse dissatisfaction withpain and agitation management, de-creased markedly after intervention, asexpected (Fig. 4). However, treatmentsshould be ordered with caution, adaptedto the minimal efficient dose, and stoppedif unnecessary. This study constitutes animprovement project in quality and safetyin health care (22, 23). Importance of thisproject is highlighted by the high inci-dence of pain (63%) and agitation (29%)in the control group. The decrease of thestress response associated with pain andagitation, characterized by tachycardia,increased myocardial oxygen consump-tion, hypercoagulability, immunosuppres-sion, and persistent catabolism (24, 25),could explain in part the better outcomeassociated with the intervention group.Actually, it was suggested that the com-bined use of analgesics and sedativescould ameliorate the stress response incritically ill patients (26, 27).

However, decrease of duration of me-chanical ventilation and rate of nosoco-

mial infections should be explained espe-cially by physician education, whichencouraged them to stop or decrease an-algesics or psychoactive drug administra-tion in the absence of pain or agitation(Appendix 1). Implementation of a sys-tematic evaluation of pain and agitationallowed providers to screen for the ab-sence of pain or agitation and therefore todecrease or discontinue administration ofanalgesic or psychoactive drugs duringthe two daily rounds of physicians.

There were few differences betweenthe groups with regard to the types ofmedications used. However, higher dosesof morphine, clonidine, and phloroguci-nol were used in the intervention group.

The number of therapeutic changes inthe way of an escalation but also in theway of a de-escalation in many analgesicand psychoactive drugs increased duringthe intervention phase to better meet pa-tients’ needs. These findings are similarto those observed after introduction of asedation algorithm based on a semiquan-titative scale (28–30). There were severaldifferences between our study and theseprevious studies. First, all patients whostayed �24 hrs in our ICU were includedin the present study, contrary to previousstudies that included only sedated andmechanically ventilated patients. This isthe first study to our knowledge that eval-uated the impact of implementation of

Table 3. Consequences and outcomes

Control Group(n � 100)

Intervention Group(n � 130) p Value

Use of an intravenous continuous sedationUse of a hypnotic, n (%) 54 (54) 72 (55) .83Duration of infusion, hrs 84 (39–201) 48 (24–144) .03Use of a morphinic, n (%) 47 (47) 60 (46) .90Duration of infusion, hrs 96 (48–216) 60 (24–144) .02Total duration of ventilation, hrs 120 (48–312) 65 (24–192) .01Number of reintubations, n (%) 6 (10) 7 (7) .77Surgical reintervention, n (%) 6 (10) 7 (9) .77Upper gastrointestinal hemorrhage, n (%) 3 (3) 2 (2) .65Thromboembolic event, n (%) 4 (4) 3 (2) .47Myocardial ischemia, n (%) 12 (12) 9 (7) .25First day of intestinal transit, days 5.0 (4.0–10.0) 6.0 (3.0–8.0) .19Maximal amount of daily gastric residuals, mL 400 (200–600) 300 (100–500) .09Intestinal occlusion, n (%) 1 (1) 2 (2) �.99Self-extubation, n (%) 3 (4) 2 (2) .65Self-removal of central venous catheter, n (%) 2 (2) 3 (3) �.99Self-removal of bladder catheter, n (%) 4 (5) 1 (1) .17Self-removal of gastric tube, n (%) 9 (12) 10 (10) .81Length of ICU stay 8.5 (4.0–14.7) 7.0 (4.0–13.0) .38Mortality rate, n (%) 12 (12) 19 (15) .76

ICU, intensive care unit.Quantitative data were expressed as median (25th–75th percentiles).

Table 4. Nosocomial infections

Control Group(n � 100)

Intervention Group(n � 130) p Value

Nosocomial infections 17/100 (17) 11/130 (8) �.05Pneumonia in ventilated patients

No. of patients, (%) 9/66 (14) 8/94 (9) .31No. of events for 1000 ventilation days 13.6 12.7 .03

Central catheter colonizationNo. of patients, (%) 7/88 (8) 3/113 (3) .10No. of events for 1000 catheterization days 6.1 2.4 �.0001

Urinary infection on catheterNo. of patients, (%) 4/83 (5) 2/115 (2) .40No. of events for 1000 bladder catheterization

days3.9 1.7 �.0001

BacteremiaNo. of patients, (%) 5/100 (5) 4/130 (3) .50No. of events for 1000 days of ICU stay 3.9 2.8 �.01

ICU, intensive care unit.Quantitative data were expressed as median (25th–75th percentiles).

1695Crit Care Med 2006 Vol. 34, No. 6

the RASS, which is the only agitationscale validated in a large population ofventilated and nonventilated and sedatedand nonsedated ICU patients. Second,this is the first study to our knowledgethat evaluated the impact of implementa-tion of specific tools allowed to evaluatepain in communicative and also in non-communicative patients.

To our knowledge, there are no pub-lished data regarding a systematic assess-ment of pain and agitation levels in orderto improve pain and agitation manage-ment in the ICU. In addition, the inci-dence of these events is not well known inICU patients due to a paucity of precise

published data (7). The incidence of painhas been reported during a painful pro-cedure (31–33) but surprisingly, thereare no data published concerning inci-dence of pain at rest. A reported inci-dence of pain and bothersome psycholog-ical memories of 30 –50% after ICUdischarge (34, 35) suggests the potentialfor improved symptom management,which could contribute to a less stressfulICU stay and improved patient outcomes.Impact of these events on the posttrau-matic stress disorder occurrence is likely(36). Recent studies that evaluated thequality of life after an ICU stay (37, 38)showed that 44% of patients had more

discomfort or pain 18 months after ICUdischarge than before ICU admission (37)and that pain was worse 6 yrs after ICUdischarge compared with general popula-tion controls (38). In addition, the level ofpain experienced in hospital is stronglyassociated with the level of pain experi-enced 6 months after discharge amongseriously ill patients (39). There are veryfew data available about agitation in theICU. An incidence of 16% of severe agi-tation defined as two or more Motor Ac-tivity Assessment Scale levels �4 in a24-hr period and sedative and/or narcoticdoses above a established sedation andanalgesia protocol or a combination oftwo or more sedatives has been reportedin a ventilated medical ICU population(40). Screening for agitation by readingthe nurses’ notes, we reported in a previ-ous study an agitation incidence of 52%in a ventilated/nonventilated medical-surgical ICU population (41). The lower in-cidence of agitation in the control group ofthe present study (29%) could be explainedby a different definition of incidence andagitation but also by an awareness of thestaff between the two studies (3 yrs). Agita-tion in ICU was associated with adverseoutcomes including longer duration of me-chanical ventilation (40), prolonged stay(40, 41), nosocomial infections (41), andunplanned extubations (40, 41).

Our study has several limitations. First,patients were not randomized. However,study was performed using a sequentialstudy of two closely related time periodsin which all consecutive patients werescreened for enrollment in the study.This design was appropriate to test thestudy hypothesis because it was essentialto prevent contamination of groupsthrough use of systematic objective eval-uation of pain and agitation in the con-trol patients (42). Second, the observa-tion rate of the evaluation of pain andagitation was significantly higher in thesecond phase. Because of a student-to-patient ratio lower than the nurse-to-patient ratio (1–2 to 12 vs. 1 to 3), stu-dents could not evaluate all the patientsbecause of an occasional high charge ofwork during the control phase. This re-inforced the results because the inci-dence of pain and agitation was lower inthe second group. Moreover, the medianobservation rate of pain was lower thanagitation. Other pain evaluation toolsshould be developed for nonventilatedICU patients who are unable to commu-nicate. Third, the incidence and intensityof pain and agitation events were ob-

Figure 3. Nurse satisfaction with pain and agitation management. This figure shows the improvementof nurse satisfaction with pain and agitation management after intervention phase in the intensive careunit (ICU). Nurses were considered satisfied if they answered that management was good or very goodin the ICU.

Table 5. Therapeutic escalation and de-escalation

No. of Therapeutic Changes per Day of Stay in ICUfor Ten Treated Patients

ControlGroup

InterventionGroup p Value

EscalationHypnotics 5.0 (3.3–8.3) 5.0 (2.5–7.6) .54Neuroleptics and clonidine 1.1 (0.5–2.0) 1.7 (1.1–4.3) .02Anxiolytics, sleep-inducing drugs, and hydroxyzine 2.2 (1.3–4.4) 2.3 (1.4–4.0) .76First WHO level analgesics 3.3 (2.1–5.0) 3.7 (2.0–6.1) .45Second WHO level analgesics 1.6 (0.9–3.0) 2.2 (1.2–3.6) .18Third WHO level analgesics 1.2 (0.5–2.5) 2.2 (1.1–5.9) .02Other analgesicsa 1.0 (0.6–1.8) 2.0 (1.2–3.8) .002

De-escalationHypnotics 2.0 (1.3–3.3) 2.5 (1.4–5.0) .12Neuroleptics and clonidine 0.8 (0.3–1.6) 1.6 (0.8–3.0) .04Anxiolytics, sleep-inducing drugs, and hydroxizine 1.0 (0.6–2.0) 1.2 (0.7–2.0) .54First WHO level analgesics 1.2 (0.8–2.3) 2.0 (1.1–2.8) .11Second WHO level analgesics 1.5 (0.7–2.0) 1.4 (0.9–2.0) .87Third WHO level analgesics 0.5 (0.2–0.9) 1.1 (0.9–1.9) .06Other analgesicsa 0.3 (0.2–1.0) 1.7 (1.0–2.5) �.0001

ICU, intensive care unit; WHO, World Health Organization.aNonsteroid anti-inflammatory, spasmolytic intestinal, and centrally acting muscle relaxant drugs.

Quantitative data expressed as median (25th–75th percentiles).

1696 Crit Care Med 2006 Vol. 34, No. 6

served only at rest. A prolonged presenceof independent observers during the dayand night was not possible during thefirst phase. We may have missed addi-tional episodes of pain and agitation dur-ing these periods.

Fourth, clonidine was ordered in caseof agitation, but its analgesic effect (43)was not evaluated. Fifth, we did not usespecific ICU tools to assess the neuropsy-chological compound of pain and agita-tion (12), such as the recently validatedConfusion Assessment Method for theICU (44). This tool, which assists in thedetermination of delirium by nurses, hasnot yet been validated in a French ICU,where the average nurse to patient ratiois 1:3. Finally, we did not contact thepatients after ICU discharge to determinewhether there were improvements inpost-ICU quality of life and rehabilitationrelative to the data reported by Desbiensand the SUPPORT group (34, 39).

CONCLUSIONS

Systematic evaluation of pain and ag-itation levels by nurses using the BPS,the ICU-adapted large size NRS, and theRASS following by a medical interventionafter initial nurse and physician educa-tion on pain and agitation management isassociated with a decrease in the inci-dence and intensity of these events inICU. In addition, this improvement pro-cess in quality and safety is associatedwith a decrease in duration of sedation,

duration of ventilation, and nosocomialinfections rate. These results could beexplained by a better match between an-algesic and psychoactive drugs adminis-tered and patients’ requirements.

ACKNOWLEDGMENTS

We are grateful for the enthusiastic sup-port of the nurses and doctors of the ICU(DAR B) at Saint Eloi Montpellier Univer-sity Hospital and the communication de-partment of Montpellier University Hospi-tal. We thank Pr. Jean-François Payen, MD,PhD, and Pr. Peter Dodek, MD, MHSc, forhelpful comments on this manuscript.

REFERENCES

1. Recommendations for sedation, analgesiaand curarization. Short text. French Societyof Anesthesia and Intensive Care. Ann FrAnesth Reanim 2000; 19:98–105

2. Jacobi J, Fraser GL, Coursin DB, et al: Clin-ical practice guidelines for the sustained useof sedatives and analgesics in the critically illadult. Crit Care Med 2002; 30:119–141

3. Joint Commission on Accreditation ofHealthcare Organization. Standards and In-tents for Sedation and Anesthesia Care. Com-prehensive Accreditation Manual for Hospi-tals. Oakbrook Terrace, IL, Joint Commissionon Accreditation of Healthcare Organiza-tions, 2001

4. Fraser G, Riker R: Monitoring sedation, agi-tation, analgesia, and delirium in critically illadult patients. Crit Care Clin 2001; 17:967–987

5. Soliman HM, Melot C, Vincent JL: Sedative

and analgesic practice in the intensive careunit: The results of a European survey. Br JAnaesth 2001; 87:186–192

6. Ramsay M, Savege T, Simpson B: Controlledsedation with alphaxalone/alphadolone. BMJ1974; 2:656–659

7. Management of the agitated intensive careunit patient. Crit Care Med 2002; 30:S97–S125

8. Kinney MR, Kirchhoff KT, Puntillo KA: Chesttube removal practices in critical care unitsin the United States. Am J Crit Care 1995;4:419–424

9. Puntillo KA, Wild LR, Morris AB, et al: Prac-tices and predictors of analgesic interven-tions for adults undergoing painful proce-dures. Am J Crit Care 2002; 11:415–429

10. Hamill-Ruth RJ: Use of analgesics in the in-tensive care unit: Who says it hurts? CritCare Med 2002; 30:2597–2598

11. Freire AX, Afessa B, Cawley P, et al: Charac-teristics associated with analgesia orderingin the intensive care unit and relationshipswith outcome. Crit Care Med 2002; 30:2468–2472

12. Hamill-Ruth RJ, Marohn ML: Evaluation ofpain in the critically ill patient. Crit CareClin 1999; 15:35–54

13. De Jonghe B, Cook D, Appere-De-Vecchi C, etal: Using and understanding sedation scoringsystems: A systematic review. Intensive CareMed 2000; 26:275–285

14. Payen JF, Bru O, Bosson JL, et al: Assessingpain in critically ill sedated patients by usinga behavioral pain scale. Crit Care Med 2001;29:2258–2263

15. Sessler CN, Gosnell MS, Grap MJ, et al: TheRichmond Agitation-Sedation Scale: Validityand reliability in adult intensive care unitpatients. Am J Respir Crit Care Med 2002;166:1338–1344

16. Ely EW, Truman B, Shintani A, et al: Moni-toring sedation status over time in ICU Pa-tients: Reliability and validity of the Rich-mond Agitation-Sedation Scale (RASS).JAMA 2003; 289:2983–2991

17. Chanques G, Jaber S, Barbotte E, et al: Val-idation of the French translated Richmondvigilance–agitation scale. Ann Fr Anesth Re-anim 2006; In Press

18. O’Connell N, Humphreys H: Intensive careunit design and environmental factors in theacquisition of infection. J Hosp Infect 2000;45:255–262

19. Legall J-R, Lemeshow S, Saulnier F: NewSimplified Acute Physiology Score (SAPS II)based on a European/North American Multi-center Study. JAMA 1993; 270:2957–2963

20. Brun-Buisson C, Abrouk F, Legrand P, et al:Diagnosis of central venous catheter-relatedsepsis. Critical level of quantitative tip cul-tures. Arch Intern Med 1987; 147:873–877

21. Girou E, Stephan F, Novara A, et al: Riskfactors and outcome of nosocomial infec-tions: Results of a matched case-controlstudy of ICU patients. Am J Respir Crit CareMed 1998; 157:1151–1158

22. Schweickert W, Gehlbach B, Pohlman A, et

Figure 4. Contributing factors of nurse dissatisfaction with pain and agitation management. Thisfigure shows the number of answers representing the first or second contributing factors of nursedissatisfaction with pain and agitation management in the intensive care unit. Medical interventionwaiting and absence of pain and agitation assessment tools were the main contributing factors ofdissatisfaction before the intervention phase and decreased markedly after.

1697Crit Care Med 2006 Vol. 34, No. 6

al: Daily interruption of sedative infusionsand complications of critical illness in me-chanically ventilated patients. Crit Care Med2004; 32:1272–1276

23. Dodek P, Keenan S, Cook D, et al: Evidence-based clinical practice guideline for the pre-vention of ventilator-associated pneumonia.Ann Intern Med 2004; 141:305–313

24. Epstein J, Breslow M: The stress response ofcritical illness. Crit Care Clin 1999; 15:17–33

25. Lewis KS, Whipple JK, Michael KA, et al:Effect of analgesic treatment on the physio-logical consequences of acute pain. Am JHosp Pharm 1994; 51:1539–1554

26. Mangano D, Siliciano D, Hollenberg M, et al:Postoperative myocardial ischemia. Thera-peutic trials using intensive analgesia follow-ing surgery. The Study of Perioperative Isch-emia (SPI) Research Group. Anesthesiology1992; 76:342–353

27. Parker S, Breslow M, Frank S, et al: Cate-cholamine and cortisol responses to lowerextremity revascularization: Correlation withoutcome variables. Perioperative IschemiaRandomized Anesthesia Trial Study Group.Crit Care Med 1995; 23:1954–1961

28. Brook AD, Ahrens TS, Schaiff R, et al: Effectof a nursing-implemented sedation protocolon the duration of mechanical ventilation.Crit Care Med 1999; 27:2609–2615

29. Brattebø G, Hofoss D, Flaatten H, et al: Effectof a scoring system and protocol for sedationon duration of patients’ need for ventilatorsupport in a surgical intensive care unit. BMJ2002; 324:1386–1389

30. De Jonghe B, Bastuji-Garin S, Fangio P, et al:Sedation algorithm in critically ill patientswithout acute brain injury. Crit Care Med2005; 33:120–127

31. Puntillo KA: Dimensions of procedural painand its analgesic management in critically illsurgical patients. Am J Crit Care 1994;3:116–122

32. Stanik-Hutt JA, Soeken KL, Belcher AE, et al:Pain experiences of traumatically injured pa-tients in a critical care setting. Am J CritCare 2001; 10:252–259

33. Gacouin A, Camus C, Le Tulzo Y, et al: As-sessment of peri-extubation pain by visualanalogue scale in the adult intensive careunit: A prospective observational study. In-tensive Care Med 2004; 30:1340–1347

34. Desbiens NA, Wu AW, Broste SK, et al: Painand satisfaction with pain control in seri-ously ill hospitalized adults: Findings fromthe SUPPORT research investigations. Forthe SUPPORT investigators. Study to Under-stand Prognoses and Preferences for Out-comes and Risks of Treatment. Crit Care Med1996; 24:1953–1961

35. Rotondi A, Chelluri L, Sirio C, et al: Patients’recollections of stressful experiences whilereceiving prolonged mechanical ventilationin an intensive care unit. Crit Care Med2002; 30:746–752

36. Jones C, Griffiths RD, Humphris G, et al:Memory, delusions, and the development ofacute posttraumatic stress disorder-related

symptoms after intensive care. Crit Care Med2001; 29:573–580

37. García Lizana F, Peres Bota D, De Cubber M,et al: Long-term outcome in ICU patients:What about quality of life? Intensive CareMed 2003; 29:1286–1293

38. Kaarlola A, Pettilä V, Kekki P: Quality of lifesix years after intensive care. Intensive CareMed 2003; 29:1294–1299

39. Desbiens NA, Wu AW, Alzola C, et al: Painduring hospitalization is associated with con-tinued pain six months later in survivors ofserious illness. The SUPPORT Investigators.Study to Understand Prognoses and Prefer-ences for Outcomes and Risks of Treatments.Am J Med 1997; 102:269–276

40. Woods J, Mion L, Connor J, et al: Severeagitation among ventilated medical intensivecare unit patients: Frequency, characteristicsand outcomes. Intensive Care Med 2004; 30:1066–1072

41. Jaber S, Chanques G, Altairac C, et al: Aprospective study of agitation in a medical-surgical ICU: Incidence, risk factors, and out-comes. Chest 2005; 128:2749–2757

42. Benson K, Hartz A: A comparison of obser-vational studies and randomized, controlledtrials. N Engl J Med 2000; 342:1878–1886

43. Khasar S, Green P, Chou B, et al: Peripheralnociceptive effects of alpha 2-adrenergic re-ceptor agonists in the rat. Neuroscience1995; 66:427–432

44. Ely EW, Margolin R, Francis J, et al: Evalu-ation of delirium in critically ill patients:Validation of the Confusion AssessmentMethod for the Intensive Care Unit (CAM-ICU). Crit Care Med 2001; 29:1370–1379

APPENDIX 1

Physicians and residents received anoral education and a written support toguide the ordering of analgesics and psy-choactive drugs during the interventionphase. They were encouraged to vigor-ously treat pain and/or agitation accord-ing to the risk-benefit ratio of analgesicand/or psychoactive drug administration.Inflammatory pain was defined by paininduced by a damage that was associatedusually with inflammation, such as osfractures, pancreatitis, and arthritis.

BPS �5 or NRS �3. Consider treatinga pain event.

1. First, diagnose the source of pain:Check for a serious event: myocardialinfarction, thromboembolic event, il-eus, peritonitis.

2. Choose the appropriate analgesic

a. Use the World health Organization(WHO) analgesic scale to treat anociceptive pain.

First WHO level: acetaminophenand/or nefopam (nefopam is a non-

opioid antinociceptive compoundthat inhibits monoamine reuptake)(41).

Second WHO level: dextropropoxy-fen or tramadol or nalbuphin.

Third WHO level: morphine or fen-tanyl or patient-controlled analge-sia.

b. Use a spasmolytic intestinal drug totreat an intestinal spasm.

c. Use centrally acting muscle relax-ants to treat a muscle contracture.

d. Use anti-inflammatory nonsteroiddrugs (AINS) to treat an inflamma-tory pain if first and second WHOlevel analgesics are inefficient.

3. Consider the risk-benefit ratio of an-algesic administration.

Presence or risk of respiratory failureor inefficient cough: second and thirdWHO levels analgesics, centrally act-ing muscle relaxants.

Presence or risk of cardiovascular fail-ure: second and third WHO level an-algesics.

Presence or risk of coronary pathol-ogy: nefopam (tachycardia).

Presence or risk of hepatic failure:acetaminophen, consider accumula-tion of analgesics.

Presence or risk of renal failure:AINS, consider accumulation of anal-gesics.

Presence or risk of peptic ulcer dis-ease: AINS.

Presence or risk of ileus: second andthird WHO level analgesics, spasmo-lytic intestinal drugs.

RASS �1. Consider treating an agita-tion event.

1. First, diagnose the source of agitation:Check for a serious event: brain in-jury, fever (sepsis), hydroelectrolyticdisorders (sodium).

2. Choose the appropriate drug.

a. Is the patient in pain? See previous.

b. Is the patient talking about anxi-ety? Use benzodiazepine.

c. Is the patient confused and/or de-scribing hallucinations?

Consider an analgesic-associated

1698 Crit Care Med 2006 Vol. 34, No. 6

adverse event: nefopam, second andthird WHO level analgesics.

Use a neuroleptic and/or clonidineif neuroleptics are inefficient orproscribed.

d. Consider withdrawal syndrome ifpatient is a psychoactive and/or athird WHO level drug user or re-ceived continuous sedation in thelast 48 hrs: test reintroduction ofthe drug, consider clonidine towithdraw previous treatment.

e. If physician is unable to diagnosethe cause of agitation because thepatient is unable to communicate,including no answer to the ques-tion “Are you anxious?”:

Is the patient ventilated? Start orincrease intravenous continuoussedation.

Is the patient not ventilated or in a

weaning course? Use a neurolepticand/or clonidine.

3. Consider the risk-benefit ratio of psy-choactive drug administration.

Presence or risk of respiratory failureor inefficient cough: benzodiazepines.

Presence or risk of cardiovascular fail-ure : clonidine, benzodiazepines, neu-roleptics.

Presence of a long QT segment onelectrocardiogram: neuroleptics.

Presence of hepatic failure: accumu-lation of clonidine, benzodiazepines,neuroleptics.

Presence or risk of ileus: benzodiaz-epines, neuroleptics.

Presence or risk of Parkinson’s dis-ease: neuroleptics.

BPS �6 or NRS �4, RASS �2. Con-sider decreasing or stopping drug admin-istration.

APPENDIX 2

Laminated plastic posters with its 0-to 10-point intensive care unit (ICU)-adapted numerical rating scale. Laminatedplastic posters placed on the wall of eachroom, showing the Behavioral Pain Scale(BPS) (14), the 0- to 10-point NumericalRating Scale (NRS) (12), and the RichmondAgitation Sedation Scale (RASS) (15, 16).The 0- to 10-point NRS (12) was adapted forICU patients who often have difficulty read-ing small print—therefore the scale wasenlarged to 3.9 � 11.8 inches. This wash-able plastic tool was approved by the infec-tion control Montpellier University Hospi-tal according to usual guidelines (18) andplaced in a plastic pocket on the poster ineach ICU room.

1699Crit Care Med 2006 Vol. 34, No. 6