The majority of neonates admitted for neonatal intensive care will require some form of respiratory support. Mechanical ventilation is reported to be a significant stressor by articulate adults and children (Novaes et al. 1997, 1999; Simini 1999; Gelinas et al. 2004) and has been associated with significant increases in biochemical stress responses that can be reduced by analgesia (Guinsburg et al. 1998). Appropriate analgesia also improves respiratory function, particularly in larger term neonates in whom decreased lung compliance, hypoxaemia and atelectasis appear to be related to pain (Bolivar et al. 1995). In preterm infants, the use of opioids improves synchrony with the ventilator and reduces the duration of oxygen therapy (Dyke et al. 1995).

Whereas adequate analgesia is regarded as essential and routine therapy for adults and older children undergoing mechanical ventilation, this is not the case in neonatal medicine, and the routine use of analgesia remains controversial. Lack of consensus among clinicians relates to differences in interpretation of the balance between risks and benefits associated with the pharmacological and non-pharmacological management of pain.

A number of clinical research studies have sought to clarify the role of routine continuous or intermittent administration of analgesic drugs (Anand et al. 2004; Simons et al. 2005; Dyke et al. 1995; Orsini et al. 1996; Guinsburg et al. 1998; Quinn et al. 1993; Saarenmaa et al. 1999). Many have focused on the preterm population, the largest of these being the NEOPAIN (NEurological Outcomes and Pre-emptive Analgesia In Preterm ventilated Neonates) Multicenter Trial (Anand et al. 2004), which enrolled 898 infants to receive continuous infusion of morphine or placebo and failed to show any improvement in short-term neurological outcomes with routine analgesia. A Cochrane systematic review on the use of opioids for term and preterm neonates receiving mechanical ventilation found significant heterogeneity between studies, particularly with respect to methods of pain assessment (Bellu et al. 2008). In most studies, opioids were started at the onset of mechanical ventilation, and the measured effects were small and inconsistent. It was not clear whether this was due to variation in measurement or infant responses. The authors concluded that there was insufficient evidence to support the routine use of opioid analgesia in all ventilated neonates but recommended pharmacological intervention in selected populations and clinical situations. Additional studies to determine clearly the efficacy, safety and long-term effects of analgesic therapies in neonates are much needed.

Given the available evidence and whilst awaiting further research and follow-up, the judicious use of pharmacological measures seems reasonable from a clinical and humanitarian perspective, particularly for selected ventilated babies. Valid indications for analgesia are:

Non-pharmacological measures including environmental and developmental interventions, such as positioning, swaddling and minimal handling, can be effective in reducing chronic distress in babies and may reduce the need for drugs in most cases (Golianu et al. 2007; Cignacco et al. 2010; Westrup et al. 2007). Although the evidence for their use in this group is less convincing than for older children and adults, opioids remain the most extensively studied and used analgesic drugs in ventilated neonates; morphine and fentanyl are the most commonly used drugs. Each drug has potentially beneficial and adverse effects, and there is no consensus as to which is optimal. Choice of drug depends to a large extent on the personal preferences, experience and familiarity of clinicians.

Assessment of pain in preverbal neonates is always challenging. Although more than 40 clinical tools are now available to facilitate the assessment of acute pain, there is a relative lack of robust and validated tools designed for the assessment of ongoing or “chronic” pain/distress that might accompany mechanical ventilation. To date, there are only three such tools, the EDIN (Echelle Douleur Inconfort Nouveau-Né) (Debillon et al. 2001), the N-PASS (Neonatal Pain, Agitation and Sedation Scale) (Hummel et al. 2008) and the COMFORTneo (van Dijk et al. 2009). All three are multidimensional scores, including physiological indicators as well as facial expressions, the most sensitive indicator of pain. However, physiological indicators lack specificity for pain, and different external methods of securing endotracheal tubes may distort or obscure facial features, which potentially limits their use in ventilated babies. In addition, it is difficult to differentiate between pain and agitation and the effects of analgesia versus sedation, especially in critically ill infants in whom severity of illness may further complicate their assessment.

Research to identify novel methods of detecting and quantifying chronic pain in neonates is ongoing. Investigators continue to build on the knowledge of behavioural responses (Holsti and Grunau 2007; Boyle et al. 2006). It is likely that the application of noninvasive techniques at the cot side, such as near-infrared spectroscopy, may provide greater insight into more specific pain indicators (Bartocci et al. 2006; Slater et al. 2008).

Opioid dependence is a well-documented phenomenon and is recognised in neonates who are exposed to prolonged therapy. Opioid withdrawal (neonatal abstinence syndrome) was first described in relation to infants exposed to prenatal opiates, but the clinical signs seen in infants following discontinuation of therapeutic opioids are similar. These include crying, sleeplessness, hypertonia, seizures, fever, sneezing, loose stools, tremors, sweating and tachypnoea. In infants receiving ECMO support, signs of withdrawal were observed in 57 % of the babies after a relatively short duration of treatment, with rapid weaning as tolerated to allow reintroduction of enteral feeds. A number of scoring tools have been used to monitor withdrawal, mostly based on the Neonatal Abstinence Score (Finnegan et al. 1975), but few of these were designed and validated for ventilated term or preterm babies. Of these, the 12-item Withdrawal Assessment Tool-1 (WAT-1) appears most promising due to its empirical development, ease of use at the patient’s bedside and psychometric properties showing high sensitivity (0.87), specificity (0.88) and excellent convergent and construct validity (Franck et al. 2008).

Work confined to the neonatal population is very limited; there are no validated weaning regimens and no prospective studies comparing various weaning protocols (Anand et al. 2010). Studies have suggested weaning by daily dose reduction with some recommending ongoing daily reductions of 5–10 % of the peak dose and others using initial larger reductions of 20–40 % followed by smaller reductions at intervals ranging from 4 to 24 hourly (Cho et al. 2007). Some have used substitution with longer-acting drugs such as methadone to facilitate weaning (Berens et al. 2006), but this approach has not been evaluated systematically in neonates. Newer scoring systems to aid the weaning of analgesic and sedative medication are not yet widely used (Franck et al. 2008; Ista et al. 2007). Much research is needed to examine the mechanisms underlying opioid tolerance in neonates and the efficacy of novel therapies to prevent opioid tolerance and withdrawal (Anand et al. 2010).

Weaning from opioid therapy will usually be indicated in the recovering or maturing neonate in whom discontinuation of mechanical ventilation is expected. Reduction of the dose of opioid analgesia aims to facilitate extubation whilst minimising the effects of withdrawal. Until further research clarifies optimal strategies, weaning should be combined with environmental measures to provide comfort and tailored to individual babies based on the duration of their therapy and signs of withdrawal phenomena.

In addition to the problems of dependence and withdrawal, a number of other adverse effects occur with the use of opioids. Results from the NEOPAIN trial showed an association between morphine and a significant increase in the duration of ventilation (Bhandari et al. 2005). Hypotension is a well-recognised unwanted effect of opioids, occurring more commonly with morphine than fentanyl. Morphine contributed significantly to hypotension observed in the first 24 h of life in babies in the NEOPAIN trial, particularly in those with pre-existing hypotension and the most immature infants (Anand et al. 2004; Hall et al. 2005). Studies have consistently shown that opioid use is significantly associated with reduced gut motility and an increase in the time taken to establish enteral feeding (Bellu et al. 2008; Menon et al. 2008; Saarenmaa et al. 1999).

Although tolerance can occur with morphine, it is seen more commonly with fentanyl, necessitating increasing doses of the drug over time to maintain effective analgesia. Few studies have been conducted in neonates, but a retrospective review of 37 term infants during extracorporeal membrane oxygenation (ECMO) showed rapid development of tolerance and requirement for high doses by 6 days of treatment (Arnold et al. 1990). The most worrying adverse effect of fentanyl is life-threatening chest wall rigidity that can occur with rapid intravenous administration of the drug, making ventilation difficult (Fahnenstich et al. 2000).

Opioids are commonly used for neonates requiring mechanical ventilation, not only for their analgesic but also for their sedative effects. Issues associated with opioid use are discussed in Sect. 38.1 above.

Benzodiazepines can be given intermittently or as continuous infusions to provide sedative, anxiolytic and hypnotic effects, but they have no analgesic effect. Although there is limited evidence to support the use of benzodiazepines, these drugs are extensively used in neonatal intensive care.

The use of benzodiazepines was introduced empirically into the care of critically ill newborn infants with the intention of reducing stress associated with mechanical ventilation and increasing the effectiveness of ventilation. However, rigorous investigation of the safety and effectiveness of this group of drugs in the neonatal population is still lacking. Challenges in measuring levels of sedation further complicate the use of benzodiazepines. Since benzodiazepines have most often been used in conjunction with opioid analgesia, it is difficult to study the effects of these drugs in isolation. Midazolam is widely used, but there have been only three high-quality randomised controlled trials in neonates (Jacqz-Aigrain et al. 1994; Anand et al. 1999; Arya and Ramji 2001). A Cochrane systematic review (Ng et al. 2003) including these trials suggested an increase in adverse neurological outcomes occurs with midazolam use, which was supported by data from other non-randomised studies (Bergman et al. 1991; Magny et al. 1994; Ng et al. 2002). The Cochrane review concluded that there were insufficient data to recommend the use of intravenous midazolam as a sedative in neonates undergoing intensive care and raised concerns about its safety. There are currently no indications for the routine use of midazolam for sedation in neonatal intensive care. Indeed, the Cochrane review of opioids for neonates receiving mechanical ventilation also concludes that, if sedation is required, morphine is safer than midazolam (Bellu et al. 2008).

Benzodiazepines produce their effect by activating GABA_A-benzodiazepine receptors, which inhibit neuronal activity. Immature GABA_A receptors in the neonate, however, have the opposite effects, increasing nociceptive behaviour and lacking sedative effects (Koch et al. 2008). Metabolism and excretion are delayed in neonates, particularly in the most immature, compared with older individuals, and prolonged sedation can occur. The optimum dose is not known, but most clinicians have used midazolam infusions of 20–60 μg/kg/h to achieve sedation. There is great variability between individuals in response to midazolam; thus dosing should be individualised if the drug is to be used (Hall et al. 2007).

The need for sedation is variable between patients, with some requiring deep sedation to tolerate ventilation, whilst others may require only light sedation. It is reasonable to suppose that such variation also exists in preverbal neonates, and this inevitably contributes to difficulties in assessment of adequacy. Since indicators of stress and pain are so similar in neonates, difficulties exist in quantification of pain relief and sedation, and it is a formidable challenge to attempt to separate the two. The N-PASS (Hummel et al. 2008) seeks to combine the assessment of prolonged pain with that of sedation, but validity and reliability have been assessed in a relatively small and heterogeneous group of infants. Another assessment tool, the State Behavioral Scale, was proposed for use in ventilated infants and young children, but this too was tested in a group with widely varying ages from 6 weeks to 6 years (Curley et al. 2006).

There is currently no consensus as to the optimal level of sedation to be achieved in ventilated infants and no well-validated scale to facilitate assessment of sedation. Further work in this area is needed before any recommendations can be made.

Discontinuation of benzodiazepines leads to disinhibition of the central nervous system due to reduced efficacy of receptor availability of GABA. Withdrawal is characterised by stimulation of the central nervous system, sympathetic nervous system activation and gastrointestinal disturbance (Birchley 2009). Benzodiazepines are rarely used as sole means of sedation, being most often combined with opioids. The withdrawal syndromes of the two types of drug show considerable overlap and are difficult to separate. Since benzodiazepines are not recommended for routine use in neonates, opioids are most likely to be employed for their sedative effects. Weaning of these drugs has been discussed in Sect. 38.1.4.

Adverse effects of midazolam are most frequently associated with intermittent bolus dosing rather than the use of continuous infusion. It causes hypotension due to vasodilatation, particularly in infants who already have low blood pressure and, if used in combination with opioids, this effect may be exacerbated (Burtin et al. 1991). Researchers have shown that decreased cerebral blood flow velocity and cerebral blood oxygenation decreases within a short time of administering a bolus of midazolam (van Alfen-van der Velden et al. 2006). Abnormal clinical neurological manifestations that may be related to cerebral hypoperfusion have been observed in a number of studies. These transient effects include hypertonia, myoclonus and abnormal movements (Bergman et al. 1991; Magny et al. 1994).