Independent risk factors for postoperative pain in need of intervention early after awakening from general anaesthesia

1. Introduction 

Despite recent advances in postoperative pain management, the proportion of patients with moderate to severe postoperative pain is still high ranging 20–80% (Harmer, 1991, Warfield and Kahn, 1995, Wickstrom et al., 2005). Apart from being a process indicator for health care quality and patient satisfaction, inadequate pain management contributes to the development of surgical wound infection (Akca et al., 1999), pulmonary (Ballantyne et al., 1998, Ellstrom et al., 1998, Nguyen et al., 2001) and cardiovascular complications (Beattie et al., 1993), chronic pain (Tasmuth et al., 1996) with long term effects on quality of life (Poobalan et al., 2001) and delirium in the elderly (Lynch et al., 1998).

Contrarily, improved perioperative pain management may not only improve patient satisfaction with the perioperative experience (Thomas et al., 1998) but also reduce length of stay in the hospital and decrease the risk of pulmonary and cardiovascular complications (Tsui et al., 1997, Kehlet and Holte, 2001, Gloth, 2001, Walder et al., 2001, Schumann et al., 2003). Pain control has been implicated in the reduction of postoperative mortality (Bonnet and Marret, 2005, Walder et al., 2001, Wu et al., 2004) although this is controversial (Liu and Wu, 2007).

At present, acute pain management programs are mostly determined by the type of surgical procedure and do not take individual patient characteristics into account (Warfield and Kahn, 1995). The desirable objectives in postoperative pain management are controversial; moreover the level that constitutes ‘acceptable’ pain remains unclear due to inconsistencies in reporting (Coll et al., 2004). It is unrealistic to expect no pain, whereas pain management programs aiming to maintain the severity of pain at levels that can be coped with would be a sensitive and important objective. Certainly this would require more than usual individual attention by well educated staff (Salmon and Manyande, 1996). Individual pain acceptance/tolerance of the patient should also be taken into account (Maroney et al., 2004).

Despite general agreement to divide acute pain into minor, moderate and severe pain, distinct cut-off points vary due to inconsistent evidence and interpretation of available data (Collins et al., 1997). In 2006 the German national guideline clearinghouse (Arbeitsgemeinschaft der Wissenschaftlichen Medizinischen Fachgesellschaften e.V., AWMF) devised a guideline for postoperative pain treatment. This suggests an intervention threshold of equal or greater than 30–50% of the maximum pain scale (Laubenthal et al., 2006). The characteristics of patients presenting with postoperative pain in need of intervention (PPINI) early after awakening from general anaesthesia have not been systematically examined.

The identification of patients at risk for PPINI would be a useful strategy to enable individual attention for their pain management (Dahmani et al., 2001). In this retrospective study, we aimed to investigate the incidence of PPINI, defined as NRS>4 at rest in the post anaesthesia care unit (PACU) early after awakening from general anaesthesia as well as to identify possible risk factors associated with PPINI.

2. Methods 

3. Results 

In total, 1736 patients were analysed in this study (Fig. 1). Out of these patients, 494 (28.5%) had a NRS>4, while 1242 (71.5%) had levels ⩽4. Patients after musculoskeletal surgery had the highest incidence of PPINI (38.2%), followed by intraabdominal surgery (33.5%), head and neck surgery (23.3%), and peripheral vascular (22.0%) and urologic surgery (18.3%). By univariate analysis there were significant differences with age, gender, ASA-PS, alcohol dependence, location of surgery, type of analgesic and type of hypnotic at induction, type of maintenance opioid, intraoperative dose of fentanyl, use of nitrous oxide, amount of intraoperative fluid application and duration of surgery (Table S1, see the online version at doi:10.1016/j.ejpain.2009.03.009). All other putative predictors had no impact on pain scores.

View Large Image Download to PowerPoint

Fig. 1. Flow diagram of data collection.

Table S1. Demographic and clinical characteristics of the patients in the study groups.

		NRS⩽4	NRS>4	p-value
		(n=1242)	(n=494)
	Age (years)	54.0 [38–66]	49.0 [36–63]	0.002
	Age (categories)			0.001
	18 age<60	754 (68.9%/60.7%)	341(31.1%/69.0%)
	Age 60	488 (76.1%/39.3%)	153 (23.9%/31.0%)
	Gender			0.001
	Female	569 (67.8%/45.8%)	270 (32.2%/54.7%)
	Male	673 (75.0%/54.2%)	224 (25.0%/45.3%)
	Height	171.7 [164–179]	170.0 [164–178]	0.284
	Weight	75.0 [65–85]	75.0 [65–86]	0.346
	BMI (kg/m2)			0.159
	Underweight (BMI⩽18)	46 (69.7%/3.7%)	20 (30.3%/4.0%)
	Normal (18<BMI⩽25)	606 (73.5%/48.8%)	219 (26.5%/44.3%)
	Overweight (25<BMI⩽30)	415 (71.4%/33.4%)	166 (28.6%/33.6%)
	Obesity (30<BMI)	175 (66.3%/14.1%)	89 (33.7%/18.0%)
	ASA-PS			<0.001
	ASA⩽2	914 (69.3%/73.6%)	405 (30.7%/82.0%)
	ASA⩾3	328 (78.7%/26.4%)	89 (21.3%/18.0%)
	Apfel’s simplified score			0.136
	2	858 (72.7%/69.1%)	323 (27.3%/65.4%)
	3	384 (69.2%/30.9%)	171 (30.8%/34.6%)
	Substance dependence
	Opioid dependence			0.460
	No	1203 (71.7%/96.9%)	475 (28.3%/96.2%)
	Yes	39 (67.2%/3.1%)	19 (32.8%/3.8%)
	Alcohol dependence			0.018
	No	1087 (70.6%/87.5%)	452 (29.4%/91.5%)
	Yes	155 (78.7%/12.5%)	42 (21.3%/8.5%)
	Nicotine dependence			0.258
	No	879 (72.3%/70.8%)	336 (27.7%/68.0%)
	Yes	363 (69.7%/29.2%)	158 (30.3%/32.0%)
	Location of surgery			<0.001
	Head and neck surgery	531 (76.7%/42.8%)	161 (23.3%/32.6%)
	Intraabdominal surgery	147 (66.5%/11.8%)	74 (33.5%/15.0%)
	Urogenital surgery	116 (81.7%/9.3%)	26 (18.3%/5.3%)
	Muskoskeletal surgery	317 (61.8%/25.5%)	196 (38.2%/39.7%)
	Peripheral surgery	131 (78.0%/10.5%)	37 (22.0%/7.5%)
	Induction hypnotic			0.027
	Propofol	954 (72.9%/76.8%)	355 (27.1%/71.9%)
	Etomidate	54 (77.1%/4.3%)	16 (22.9%/3.2%)
	Midazolam	7 (77.8%/0.6%)	2 (22.2%/0.4%)
	Thiopental	227 (65.2%/18.3%)	121 (34.8%/24.5%)
	Induction opioid			0.009
	Fentanyl	1049 (70.5%/84.5%)	439 (29.5%/88.9%)
	Alfentanil	12 (100.0%/1.0%)	0 (0.0%/0.0%)
	Remifentanil	181 (77.0%/14.6%)	54 (23.0%/10.9%)
	Sufentanil	0 (0.0%/0.0%)	1 (100.0%/0.2%)
	Use of nitrous oxide			0.001
	Yes	106 (60.9%/8.5%)	68 (39.1%/13.8%)
	No	1136 (72.7%/91.5%)	426 (27.3%/86.2%)
	Inhalation anaesthetic			0.066
	None	514 (74.8%/41.4%)	173 (25.2%/35.0%)
	Isoflurane	250 (70.0%/20.1%)	107 (30.0%/21.7%)
	Desflurane	330 (67.9%/26.6%)	156 (32.1%/31.6%)
	Sevoflurane	148 (71.8%/11.9%)	58 (28.2%/11.7%)
	Application of TIVA			0.100
	No	824 (70.3%/66.3%)	348 (29.7%/70.4%)
	Yes	418 (74.1%/33.7%)	146 (25.9%/29.6%)
	Maintenance opioid			0.010
	None	52 (80.0%/4.2%)	13 (20.0%/2.6%)
	Fentanyl	652 (68.7%/52.5%)	297 (31.3%/60.1%)
	Remifentanil	538 (74.5%/43.3%)	184 (25.5%/37.2%)
	Intraoperative dose of opioid for maintenance
	Fentanyl (mg)	0.28 [0.00–0.45]	0.40 [0.00-0.50]	<0.001
	Remifentanil (ug/kg/min)	0.08 {0.07–0.09}	0.06 {0.05–0.08}	0.082
	Mean intraoperative oxygen saturation	99 [98–100]	99 [98–100]	0.190
	Minimal intraoperative oxygen saturation	97 [96–99]	97 [96–99]	0.617
	Maximal intraoperative heart rate	80 [70–90]	80 [70–90]	0.137
	Minimal intraoperative heart rate	60 [50–65]	60 [50–65]	0.862
	Maximal intraoperative systolic BP	130 [120–145]	130 [120–150]	0.252
	Minimal intraoperative systolic BP	95 [85–100]	90 [85–100]	0.067
	Maximal intraoperative diastolic BP	74 [65–80]	75 [70–80]	0.104
	Minimal intraoperative diastolic BP	50 [40–55]	50 [40–55]	0.311
	Intraoperative crystalloid fluid application	1000 [1000–2000]	1500 [1000–2000]	<0.001
	Total intraoperative fluid application	1422 [1000–2000]	1500 [1000–2000]	<0.001
	Analgesic before emergence			0.307
	None	300 (73.5%/24.2%)	108 (26.5%/21.9%)
	Non-opioid	786 (70.7%/63.3%)	325 (29.3%/65.8%)
	Opioid	64 (78.0%/5.2%)	18 (22.0%/3.6%)
	Non-opioid and opioid	92 (68.1%/7.4%)	43 (31.9%/8.7%)
	Duration of surgery	92.5 [55–135]	110 [75–155]	<0.001

Data are median [25–75% percentiles], frequencies (percentage in row/percentage in column) or mean {95% confidence interval for mean}.

On multivariate analysis, younger age (OR=1.300 [1.007–1.678], p=0.044), female gender (OR=1.494 [1.138–1.962], p=0.004), obesity (OR=1.683 [1.226–2.310], p=0.001), use of nitrous oxide (OR=1.621 [1.110–2.366], p=0.012), longer duration of surgery (OR=1.165 [1.050–1.292], p=0.004), location of surgery (musculoskeletal OR=2.026 [1.326–3.095], p=0.001; intraabdominal OR=1.869 [1.148–3.043], p=0.012), and ASA-PS I–II (OR=1.519 [1.131–2.039], P=0.005) were identified as independent risk factors for PPINI (Table S2, see the online version at doi:10.1016/j.ejpain.2009.03.009).

Table S2. Independent risk factors of PPINI early after awakening from general anaesthesia (only significant variables).

		Regression coefficient (SE)	Odds ratio	95.0% CI for odds ratio		p
				Lower	Upper
	Younger age (age:18 age<60 vs. age 60years)	0.262 (0.130)	1.300	1.007	1.678	0.044
	Female (vs. male)	0.402 (0.139)	1.494	1.138	1.962	0.004
	BMI					0.014
	Overweight (vs. normal)	0.139 (0.127)	1.149	0.895	1.474	0.276
	Obesity (vs. normal)	0.520 (0.162)	1.683	1.226	2.310	0.001
	Underweight (vs. normal)	0.298 (0.294)	1.348	0.757	2.399	0.310
	Use of nitrous oxide (vs. without)	0.483 (0.193)	1.621	1.110	2.366	0.012
	Duration of surgery (per hour)	0.153 (0.053)	1.165	1.050	1.292	0.004
	Location of operation					<0.001
	Head and neck surgery (vs. peripheral vascular surgery)	0.090 (0.238)	1.094	0.686	1.746	0.706
	Intraabdominal (vs. peripheral vascular surgery)	0.625 (0.249)	1.869	1.148	3.043	0.012
	Urogenital (vs. peripheral vascular surgery)	−0.088 (0.303)	0.916	0.506	1.659	0.772
	Musculoskeletal (vs. peripheral vascular surgery)	0.706 (0.216)	2.026	1.326	3.095	0.001
	ASA-PS I–II (vs. III–V)	0.418 (0.150)	1.519	1.131	2.039	0.005

Variables included: age, gender, BMI, use of nitrous oxide, type of inhalation anaethesic, application of TIVA, duration of surgery, induction opioid, maintenance opioid, analgesic before emergence, location of surgery, ASA-PS, preoperative Apfel’s simplified score, opioid dependence, alcohol dependence, nicotine dependence, intraoperative fluid application, intraoperative dose of fentanyl, intraoperative dose of remifentanil.

The patients with PPINI stayed significantly longer in PACU (89.6min [70–120] vs. 80min [60-100], p<0.001) and had a longer median hospital length of stay (6.6 days [4.0–8.8] vs. 6.0 days [3.2–7.8], p<0.001) and postoperative stay (5.0 days [3.0–6.5] vs. 4.1 days [2.5–5.8], p<0.001). Patients with PPINI experienced significantly more PONV (10.3% vs. 6.2%, p=0.003), more psychomotor agitation (5.5% vs. 2.7%, p=0.004) (Table S3, see the online version at doi:10.1016/j.ejpain.2009.03.009) and received opioids or opioid in combination with non-opioid to a higher proportion for their pain management (62.8% vs.24.2% p<0.001) (Table 1). Moreover, patients with PPINI consumed more piritramid (8.0mg [5.0–12.0] vs. 5.0mg [3.0–7.8], p<0.001) in PACU than patients without PPINI (Table 2). Moreover, piritramid consumption per hour was significantly higher in younger age group (4.3mg/h [2.7–6.4mg/h] vs. 3.6mg/h [2.3–6.0mg/h], p=0.041) (Table 3).

Table 1. Pain management in PACU.

	Pain management	NRS 4	NRS>4	p-value
		(n=1242)	(n=494)
	No analgesic or non-opioid analgesia	942 (75.8%)	184 (37.2%)	<0.001
	Opioid or non-opioid+opioid	300 (24.2%)	310 (62.8%)

Data are frequencies (percentage in column).

Table 2. Opioid consumption in PACU.

		n	NRS 4	NRS>4	p-value
	Piritramid (mg)	563	5.0 [3.0–7.8]	8.0 [5.0–12.0]	<0.001
			(n=265)	(n=298)
	Morphine (mg)	33	5.0 [5.0–5.0]	5.0 [5.0–10.0]	0.166
			(n=18)	(n=15)

Data are median [25–75% percentiles].

Table 3. Opioid consumption in PACU between younger and older group.

		n	18 age<60	Age 60	p-value
	Piritramid (mg/h)	563	4.3 [2.7–6.4]	3.6 [2.3–6.0]	0.041
			(n=391)	(n=172)

Data are median [25–75% percentiles].

Table S3. Events in PACU.

		NRS 4	NRS>4	p-value
		(n=1242)	(n=494)
	PACU LOS	80 [60–100]	89.6 [70–120]	<0.001
	Maximal heart rate	83 [75–90]	85 [80–95]	<0.001
	Minimal heart rate	70 [60–80]	70 [60–75]	0.830
	Shivering			0.257
	No	1098 (88.4%)	427 (86.4%)
	Yes	144 (11.6%)	67 (13.6%)
	PONV			0.003
	No	1165 (93.8%)	443 (89.7%)
	Yes	77 (6.2%)	51 (10.3%)
	Psychomotor agitation			0.004
	No	1209 (97.3%)	467 (94.5%)
	Yes	33 (2.7%)	27 (5.5%)
	Hospital LOS	6.0 [3.2–7.8]	6.6 [4.0–8.8]	<0.001
	Postoperative LOS	4.1 [2.5–5.8]	5.0 [3.0–6.5]	<0.001

Data are median [25–75% percentiles] or frequencies (percentage in column).

4. Discussion 

The most important result in our study is that when choosing NRS>4 as the intervention limit, 28.5% of our patients experienced PPINI early after awakening from general anaesthesia. Kalkman et al. found a rate of 25.8% of severe postoperative pain defined as a NRS>7 within the first 60min after awakening from anaesthesia (Kalkman et al., 2003). Katz et al. recorded 54.1% of patients with clinically meaningful acute pain defined as a NRS⩾5 at any point within 24h postoperatively (Katz et al., 2005). Definition of the outcome parameter, period of observation as well as patient population, certainly contributed to the differences. Additionally, exclusion criteria could add to this fact, since cardiac as well as thoracic surgeries are known to cause major postoperative pain. In addition, the exclusion of surgery under the combination of general and regional anaesthesia such as major bowel surgery or joint replacement surgery should be taken into account. According to standard care in our institution the timely application of longer lasting analgesics prior to emergence most probably decreased the rate of postoperative pain early after awakening. 76.5% of our patients received some form of analgesic therapy (non-opioid [metamizol, paracetamol] and/or opioid [piritramid, morphine]) that was aimed at the postoperative period.

Univariate analysis demonstrated that median age was significantly younger in the PPINI group (49 years vs. 54 years, p<0.002). For reasons of simplification for clinical application, patients were categorized into those younger than 60 years and those equal and over 60 years of age corresponding to previous studies (Gagliese and Melzack, 2003). Multivariate analysis identified younger age as an independent risk factor for PPINI. This correlates well with other studies (Bisgaard et al., 2001, Caumo et al., 2002, Kalkman et al., 2003). Younger age was significantly associated with a high dose of analgesics consumption (Joels et al., 2003). In our study we could not demonstrate any relationship between total analgesic consumption and age. However, piritramid consumption per hour was significantly higher in younger age group.

We showed that female gender was an independent risk factor for PPINI. This is in accordance with former studies (Thomas et al., 1998, Kalkman et al., 2003). Gender-related differences in pain have been clearly shown in experimental settings. Clinical studies of such differences have produced conflicting findings. It was demonstrated that postoperative baseline pain was significantly greater in female subjects than in male subjects (Averbuch and Katzper, 2001), women reported more pain of at least moderate intensity than men (Rosseland and Stubhaug, 2004) and women experienced more severe early postoperative pain (Uchiyama et al., 2006). However, gender differences may exist only in pain perception, with females typically reporting more negative responses to pain than males (Keogh and Herdenfeldt, 2002). Analgesics may have the same or greater analgesic effect in women than in men (Gear et al., 1999). A study of 2298 Chinese patients showed that females consumed significantly less morphine via patient-controlled analgesia (PCA) in the first three postoperative days than was the case for males (Chia et al., 2002). Joels et al. demonstrated that female gender was one of the independent predictors for decreased narcotic use in patients underwent colectomy (Joels et al., 2003). This would have implications on gender specific thresholds for the management of acute postoperative pain.

Our study showed that compared to patents with a BMI>18kg/m2 and 25kg/m2, obese patients with a BMI of >30kg/m2 had a higher odds of being in the higher pain group postoperatively (OR=1.683, p=0.001). Literature examining the relationship of BMI and risk of pain is limited. Chung et al. demonstrated that higher BMI was a significant predictor of severe pain in the PACU (Chung et al., 1997). Massaron et al. reported a BMI>25kg/m2, to be an independent risk factor for higher postoperative pain following inguinal hernia repair (Massaron et al., 2007). The implications for pain management in patients with obesity remain unclear since evidence on analgesic requirements are lacking.

Longer duration of surgery (OR=1.165, p=0.004) was identified as another independent risk factor. Patients in the PPINI group had a median length 17.5min longer than those with pain NRS⩽4 (p<0.001). Chung et al. demonstrated that in a cohort of 10008 patients after ambulatory surgery, patients in the severe pain group had a 25min longer duration of anaesthesia (Chung et al., 1997). Similarly Joels et al. found a positive relationship of increased postoperative morphine consumption and operative time (Joels et al., 2003). This might suggest that the long duration of surgery correlate with the extend of surgery and plays a role in determining postoperative pain early after awakening.

In the univariate analysis we demonstrated that patients after musculoskeletal surgery had the highest incidence of pain, followed by intraabdominal surgery, head and neck surgery, and peripheral vascular and urologic surgery. Chung et al. reported a similar but lower incidence of postoperative pain after ambulatory surgery (Chung et al., 1997). In our multivariate analysis, musculoskeletal and intraabdominal surgery, were identified to be independent risk factors for PPINI. Kalkman et al. in 1416 surgical inpatients undergoing various procedures except cardiac surgery could identify type of surgery as an independent predictor for severe pain within the first postoperative hour (Kalkman et al., 2003). This implies that anaesthetists should pay additional attention to those patients who undergo a potentially painful operation.

Univariate analysis showed a higher proportion of patients with ASA-PS I–II in the PPINI group. Multivariate analysis confirmed lower ASA-PS to be an independent risk factor for PPINI (OR=1.519, p=0.005). Caumo et al. in a study of patients after abdominal surgery found patients with ASA-PS III to have more postoperative pain defined as a VAS>30mm compared to ASA-PS I at 12 and 24h postoperatively (Caumo et al., 2002). Peters et al. did not find any significant difference in ASA-PS with regards to pain defined as an average pain intensity of VAS>=40mm within the first four postoperative days (Peters et al., 2007). Whereas Chung et al. examined ambulatory surgical patients and showed that ASA-PS I patients had a higher incidence of severe pain In the PACU, which is in accordance well with our study (Chung et al., 1997). Choice of cut-off, different study design and differences in patient population may account for the different results. However, ASA-PS may have a time dependent influence on postoperative pain intensity. Testing this hypothesis in a well designed prospective study may give further evidence in this direction.

In univariate analysis we identified patients with nitrous oxide (N2O) to have experienced more PPINI compared with patients not having received N2O (13.8% vs. 8.5%, p=0.001). Multivariate analysis confirmed this differences (OR=1.621, p=0.012). It could be shown that clinical concentrations of noxious i.v. and inhalation general anaesthetics excite sensory neurons by selectively activating TRPA1, a key ion channel in the pain pathway (Matta et al., 2008). Leung et al. in a study with 228 patients did not see a significant difference in postoperative pain scores for the use of N2O (Leung et al., 2006). However due to the small number of patients receiving N2O in our population (n=174) we recommend caution in interpreting this result.

Several studies demonstrated that type of anaesthetic and choice of intraoperative analgesic regime (Aubrun et al., 2008) was important in influencing postoperative pain. In our study type of hypnotic as well as analgesic for induction and maintenance, intraoperative dose of fentanyl, alcohol dependence and intraoperative fluid application were significant in univariate but not in multivariate analysis.

Several limitations of this study might have influenced the results. Due to the retrospective design, no inference regarding a causal link between the proposed risk factors and the chosen outcome can be made with certainty. Choosing exclusion criteria for reasons of reducing the possibility of confounding factors, might have itself confounded the results. For example, the exclusion of regional anaesthesia alone or in combination with general anaesthesia might have had a considerable influence on the incidence of immediate postoperative pain, as Aubrun et al. demonstrated (Aubrun et al., 2008).

Further variables that have been reported to play a role in determining postoperative pain such as presurgical emotional status (Ozalp et al., 2003), preoperative pain threshold (Pan et al., 2006), ethnic differences (Tan et al., 2008), preoperative pain status (Kalkman et al., 2003), patient expectation (Thomas et al., 1998), chronic sleeping difficulties (Mamie et al., 2004), marital status (Katz et al., 2005), operative technique (Joels et al., 2003, Thibault et al., 2007), were not available in our study.

In conclusion, despite recent advances in postoperative pain management, the proportion of patients with PPINI was still 28.5%. On multivariate analysis, younger age, female gender, obesity, use of N2O, longer duration of surgery, musculoskeletal and intraabdominal surgery and ASA-PS I–II were identified to be independent risk factors for PPINI. The identification of patients at high risk for PPINI would be a useful strategy to enable individual attention for their pain management.