Patient-controlled analgesia (PCA) is a common, effective method for achieving postoperative pain control. Classically, opioids are self-administered intravenously using a programmable infusion pump. However, other agents and modes of delivery are available, including local anesthetics administered via peripheral nerve catheter or epidural PCA. This chapter reviews the safe use and risks and benefits of PCA, provides extensive guidance for the practical management of pain with PCA modalities, and provides a consideration of special clinical situations in which PCA may be particularly valuable. Recommendations for PCA use and monitoring of high-risk patients are also included.
Keywordsintravenous patient-controlled analgesia, patient-controlled analgesia, patient-controlled analgesia monitoring, patient-controlled analgesia safety, patient-controlled epidural analgesia
Patient-controlled analgesia (PCA) has become a standard technique in the clinical treatment of pain. PCA systems allow patients to self-administer predetermined doses of analgesic medication, and to record patient usage information during the previous 1-hour and 24-hour periods. This information can help optimize drug delivery based on the analgesic requirement and pattern of use of the individual patient.
Current PCA models usually include a combination of physician-determined variables, including the initial loading dose, demand (bolus) dose, lockout interval, basal continuous infusions, and 1- to 4-hour maximal dose limits. The patient then controls the timing of the demand dose by pressing a button to trigger administration of the drug via an intravenous line. Newer devices allow for entries in units of μg, mg, and mL, thereby reducing the potential for programming errors. Optimization of efficacy and safety depends on the selection of a demand dose large enough to provide sufficient analgesia but small enough to minimize side effects. A lockout interval is the time during which there will be no drug delivery, even if the patient pushes the demand button. Although no dose will be delivered during the lockout period, the frequency of demand presses is nonetheless collected and can be used to help assess patient comfort. Theoretically, use of a lockout interval which is less than the time to peak effect of the drug may result in inadvertent over-dosage due to stacking of analgesic doses. However, lockout intervals between 5 and 10 minutes appear optimal regardless of the opioid used.
This chapter details the risks and benefits of PCA, recommended dosing strategies for several PCA modalities, and use of PCA in special clinical populations.
Safety and Efficacy of Patient-Controlled Analgesia
Advantages of Patient-Controlled Analgesia
PCA is one of the most popular methods for providing postoperative analgesia. Data suggest use of PCA is associated with higher patient satisfaction, and enhanced perception of control over pain and recovery after surgery. In several studies, PCA pumps have been shown to be more effective than intermittent intramuscular (IM) or intravenous (IV) injections. Other advantages include avoidance of (painful) IM injections, and rapid treatment of pain due to not having to involve nurses in the administration of analgesia.
Because of the ease with which each demand dose can be provided, small boluses can be given frequently. When used properly, the ability of patients to titrate analgesics to their needs can theoretically generate a steady plasma level of analgesia and avoid the peaks and troughs associated with bolus dosing on a 3- to 4-hour basis. PCA may avoid subtherapeutic opioid concentration troughs, which can be associated with unpleasant recovery, guarding, poor chest expansion, and reluctance to mobilize. PCA may also help avoid excessive peak plasma concentrations, with associated respiratory depression and sedation.
Two early meta-analyses produced somewhat contradictory results regarding analgesic efficacy of PCA compared with IM nurse administered opioids. One indicated significant patient preference and better pain scores for PCA, compared with IM analgesia, but found no differences in total opiate consumption or postoperative length of stay. Although the other study also indicated greater patient satisfaction with PCA, there was no statistically significant difference in pain relief in patients treated by PCA and non-PCA methods. The most recent meta-analyses performed to study the efficacy and safety of PCA both concluded that overall there is moderate to low-quality evidence that PCA provides slightly better pain control and increased patient satisfaction compared with nonpatient-controlled methods, but patients used higher total doses of analgesia when treated by PCA. Side effects were similar between the groups, with the exception of pruritus, which occurred significantly more frequently in patients using PCA. Other studies report similar incidences of nausea and vomiting, sedation, pruritus, and bowel function, suggesting that the differences in total opioid dosages between PCA and non-PCA approaches may be relatively unimportant. Given the continuing popularity of PCA, these results are surprising: It appears that good results can be obtained regardless of delivery method, provided that analgesia can truly be given on demand with appropriate doses and intervals. In many circumstances, PCA is the best way to achieve these goals.
Disadvantages of Patient-Controlled Analgesia
Disadvantages of PCA can constrain its use and effectiveness. The most frequent negative perceptions relate to inadequate analgesia and/or presence of side effects, but some patients also report not trusting the PCA pump, fearing overdose or addiction, and having difficulty understanding and/or using the PCA. Chumbley et al. reported 22% of patients feared addiction and 30% feared overdose, much higher than the 4% and 11%, respectively, reported by Kluger and Owen. However, 43% of patients in the former study did not receive preoperative education about PCA, whereas all patients in the latter study received education about pain management and PCA prior to surgery. In a recent prospective study, 49% of patients reported not knowing if they would receive a dose of medication when they pushed the PCA button, and of these, 22% believed that this uncertainty exacerbated their pain. These findings were not affected by patient education about using the PCA.
Sedation and respiratory depression can occur as a result of repeated excessive PCA use, mistaking the PCA handset for the nurse-call button, and family, visitor, or unauthorized nurse-activated demand boluses. Operator errors can occur via programming of incorrect bolus dose size, concentration, background infusion, and/or unintended background infusion. Incorrect procedures may lead to use of the wrong syringe or analgesic mixture. Standardization of protocols and drug concentrations within an institution may reduce the chance of program errors. Newer PCA pumps have been designed with a human factors approach, which has significantly reduced both programming times and rates of programming errors.
Mechanical malfunctions of a PCA pump can occur, with inadvertent excessive medication delivery. Despite this, the safety of PCA is supported by meta-analysis. There is no significant difference in the frequency or severity of oversedation and respiratory depression with PCA than with conventional intramuscular or intravenous dosing.
Safety of Patient-Controlled Analgesia
Safe use of PCA requires that the patient controls the analgesic delivery. Increasing plasma concentrations of opioid usually cause sedation prior to causing clinically significant respiratory depression. Sedation usually impairs the ability of the patient to activate the PCA. Both nursing personnel and the patient’s family members must understand this concept, so that only the patient pushes the demand button. Ideally, patients, nurses, and family members should receive education about PCA use. Not every patient is a good candidate for PCA: Patients must be cooperative, comprehend the concept, and be able to push the PCA button. PCA may not be appropriate for very young children, or for patients with certain mental or physical limitations. Nurse controlled analgesia (NCA) may be used if the patient’s age, developmental level, or muscle strength interact with the ability to use the device. NCA is a safe and effective method of analgesic administration in the pediatric ICU setting. Finally, because of pharmacokinetic and pharmacodynamic variability among patients, conventional PCA settings may need to be adjusted for the individual.
Device misconnections leading to wrong-route medication administration have been the subject of at least two Sentinel Event Alerts issued by the Joint Commission. Sentinel Event Alert #36 cites nine cases of tubing misconnections involving seven adults and two infants. Deaths occurred in eight of these instances, and one resulted in permanent loss of function. These cases included instances of epidural infusions misconnected to peripheral IV lines; notably, these incidents are believed to occur more frequently than reported, since patient harm does not always result. Recommendations in the updated Sentinel Event Alert #53 focus on system redesign, rather than clinician practice changes in order to prevent misconnection errors. Specifically, the Joint Commission recommends the adoption of new connector standards in development that will make it virtually impossible to (mis)connect delivery systems with different functions.
Importance of Acute Pain Service
Implementation of an Acute Pain Service (APS), which often consists of a team of physicians and nurses that are well educated about PCA, may also promote PCA safety and efficacy. Comparison of PCA managed by an APS versus PCA managed by the surgical staff indicated that patients with APS-supervised PCA had significantly fewer side effects, were more likely to have adjustments made to the PCA dose in response to inadequate analgesia or side effects, and were more likely to transition to oral opioids rather than IM opioids after PCA. This suggests that an APS is more likely to tailor the PCA to suit individual patients. Some of the benefits ascribed to PCA may be due to an association between PCA use and supervision of the analgesic regimen by concerned and knowledgeable clinicians.
Types of Patient-Controlled Analgesia
Intravenous Patient-Controlled Analgesia
Many opioids have been used effectively for intravenous patient-controlled analgesia (IV PCA). Opioids that are pure μ-receptor agonists tend to be the first choice for IV PCA. The ideal opioid for IV PCA has a rapid onset of action, high efficacy, and intermediate duration of action without significant accumulation of drug or metabolites over time. Morphine, hydromorphone, and fentanyl most closely fulfill these criteria and are widely used for opioid-based IV PCA. Conversely, meperidine is generally considered a poor choice for IV PCA agent because the active metabolite, normeperidine, can accumulate and cause CNS excitation, including delirium, tremors, myoclonus, and seizures. However, there may be occasions when meperidine is a reasonable analgesic option. The most recent study examining the safety and efficacy profile of meperidine PCA indicated a CNS toxicity rate of 2%, and recommended a maximum safe dose of 10 mg/kg per day for no longer than 3 days. Patients should be without comorbid renal or hepatic dysfunction, and require careful evaluation and monitoring. All opioids have a similar spectrum of adverse effects, although qualitative differences are detectable. The patient’s clinical history and hospital protocols tend to influence the choice of opioid selected for IV PCA. There are few prominent differences in pain scores and incidence of adverse effects between different opioids. Consequently, patients tend to be satisfied with PCA regardless of the opioid used. The typical dosing, lockout interval, and basal infusion parameters are indicated in Table 13.1 .
|Drug||Bolus (Dose)||Lockout Interval (min)|
|Remifentanil (labor)||0.5 μg.kg −1||2|
For safety reasons, a continuous background infusion with IV PCA should only rarely be prescribed for spontaneously breathing opioid-naïve patients. Continuous infusions pose increased risk for respiratory depression. If a patient becomes sedated, continuing delivery of opioid at a basal rate may cause respiratory depression. Continuous opioid infusion in association with PCA may provide more constant plasma opioid levels and improve analgesia. However, other investigators found that addition of a basal infusion rate did not reduce pain, fatigue, or anxiety, and failed to improve quality of sleep. The number of patient demands, number of supplemental bolus doses, and total opioid use were also not changed in patients receiving basal infusions of opioids. Additionally, most PCA programming errors that have resulted in adverse side effects occurred during the use of basal infusions. In selected opioid-tolerant patients with high opioid requirements, a background infusion may be used to deliver the equivalent of the usual opioid dose taken by the patient. Use of a background rate of infusion may necessitate higher vigilance and/or increased monitoring of the patient.
The addition of ketamine (an N-methyl- d -aspartate [NMDA] receptor antagonist) to IV PCA solutions may improve analgesia in some, but not all, circumstances. The use of ketamine as a PCA adjunct is biologically attractive for several reasons: NMDA receptor activation is associated with the early development of opioid tolerance, an effect that could be ameliorated by ketamine. Additionally, NMDA receptor antagonists are inherently analgesic, providing both a (direct) pain-control benefit and an (indirect) opiate sparing effect. Optimization of postoperative IV PCA after spine and hip surgery indicated the ideal ratio of morphine and ketamine to be 1:1, with a lockout interval of 8 minutes. However, two studies showed that either ketamine as an adjunct for IV PCA did not improve pain, or that the potential usefulness of ketamine was offset by a high incidence of adverse effects and no opioid-sparing. One study used a PCA regimen comprising background infusion plus PRN bolus doses of fentanyl/ketamine/ondansetron in patients at high risk of postoperative nausea and vomiting after major spine surgery. Although total fentanyl consumption was reduced in the ketamine-PCA group, the severity of nausea was significantly higher, with similar reported pain scores. It is also important to consider the possibility that ketamine can arouse psychomimetic effects and impair cognition.
Clonidine is an α2-adrenoreceptor agonist with analgesic properties. The addition of clonidine to morphine PCA significantly reduced nausea and vomiting in a female population undergoing lower abdominal surgery. However, other studies fail to show significant benefits from inclusion of clonidine with IV PCA.
Dexmedetomidine is a potent, highly selective α2-adrenoreceptor agonist, with analgesic, anxiolytic, and sedative properties, but without effects on respiratory function. Perioperative infusion is associated with improved analgesia, reductions in opiate consumption, and less postoperative nausea and vomiting. To date, there is one study investigating the analgesic effect of adding dexmedetomidine to morphine IV PCA. Patients received either morphine 1 mg/mL PCA or morphine 1 mg/mL plus dexmedetomidine 5 μg/mL PCA, programmed to deliver 1 mL per demand bolus, with a 5-minute lockout and no background infusion. The addition of dexmedetomidine resulted in superior analgesia, significant morphine sparing, and less nausea. These benefits were achieved without additional sedation or undesired hemodynamic changes.
Nonintravenous Patient-Controlled Analgesias
The defining concept of PCA is patient-demand drug administration. Although IV PCA is the most common and studied route of delivery, the two common alternative routes are patient-controlled epidural analgesia and patient-controlled peripheral nerve catheter analgesia.
Patient-Controlled Epidural Analgesia
In many situations, epidural analgesia is superior to IV PCA. A meta-analysis demonstrated that for all types of surgery and pain assessments, all forms of epidural analgesia, including patient-controlled epidural analgesia (PCEA), provided superior postoperative analgesia compared with IV PCA. This conclusion is corroborated by a systematic review of the analgesic efficacy of epidural analgesia versus systemic opioids.
The beneficial postoperative effects of epidural analgesia are more apparent for high-risk patients or those undergoing higher risk procedures. Epidural analgesia with a local anesthetic combined with an opioid provides better postoperative analgesia than epidural or systemic opioids alone, and may improve postoperative outcome. Use of local anesthetic alone may result in excessive motor blockade. Despite numerous investigations, the ideal PCEA epidural analgesic solution or the ideal delivery variables remain controversial. In contrast to IV PCA, a continuous background infusion is routinely used for PCEA. A background infusion can maintain a continuous segmental sensory block, but may increase the incidence of hypotension and motor block. The addition of clonidine (2 μg/mL) to ropivacaine-fentanyl PCEA after total knee arthroplasty reduced the need for opioid rescue without jeopardizing hemodynamics. Similarly, the addition of clonidine (10–20 μg/h) to bupivacaine-fentanyl PCEA produced both dose-dependent improvement in analgesia at rest and dose-dependent decrease in blood pressure and pulse rate, and an increase in vasopressor requirement. PCEA with clonidine plus local anesthetic can also avoid some of the usual opioid-related side effects such as nausea or pruritus. To facilitate transition to oral analgesia, the PCEA settings can be reduced gradually rather than abruptly, terminating the PCEA. This can be done, for example, by eliminating the basal rate 6 hours prior to stopping the PCEA.
In addition to providing better pain control, epidural analgesia also has the potential benefits of decreased morbidity, such as fewer cardiopulmonary complications, less thromboembolism, better mental status, earlier restoration of gastrointestinal function, enhanced functional exercise capacity and health-related quality of life, and earlier discharge from the hospital. A large database study further suggests mortality is lower for patients who receive neuraxial or combined neuraxial-general anesthesia, and transfusion requirements are lowest for those receiving neuraxial anesthesia compared with all other groups. Cancer recurrence and metastasis may also be lower in patients who receive paravertebral or epidural analgesia compared with conventional systemic opioids after mastectomy or prostatectomy. Finally, a recent meta-analysis of 14 prospective and retrospective studies of a range of cancer types demonstrated a positive association between epidural analgesia and overall survival, but no difference in recurrence-free survival compared with general anesthesia with opioid analgesia.
The potential benefits of PCEA must be weighed against potential risks associated with placement of a catheter, which include epidural hematoma, infection, or neurological injury. In particular, thromboprophylaxis with potent anticoagulants may limit the use of PCEA. Suggested regimens for starting a PCEA by surgical indication are shown in Table 13.2 .
|Surgical Site||Drug and Concentration||Basal Rate (mL/h)||Demand Dose (mL)||Lockout (min)|
|Obstetric (Labor)||Bupivacaine 0.025% + Fentanyl 10 μg.mL −1||3||3||10|
|Obstetric (Labor)||Ropivacaine 0.08% + Fentanyl 2 μg.mL −1||5||5||10|
|1. Abdomen |
2. Lower extremity (lumbar epidural)
|Bupivacaine 0.0625% + Fentanyl 5 μg.mL −1||4||4||10|
|Hysterectomy||Bupivacaine 0.125% + Fentanyl 5 μg.mL −1 ± Clonidine 0.75 μg.mL −1 (NB: 30 mL limit/4 h)||4||3||10|
|Hip or knee surgery||Bupivacaine 0.06% + Hydromorphone 10 μg.mL −1||4||4||10|
|Hip or knee surgery||Bupivacaine 0.06% + Clonidine 1 μg.mL −1||4||4||10|
Peripheral Nerve Catheter Patient-Controlled Analgesia
Nerve block techniques are increasingly popular for management of postoperative pain, particularly after orthopedic surgery. Many common nerve blocks, including brachial plexus, sciatic, and femoral nerve blocks, are amenable to having peripheral nerve catheters inserted for extended analgesia. Peripheral nerve blockade on both the upper and lower extremities can improve postoperative analgesia and patient satisfaction. Infections and neurological complications, although rare, are possible. In contrast with neuraxial blocks, there is less concern about interaction of anticoagulants and peripheral nerve blocks. Ropivacaine may be associated with reduction of complete motor and sensory block, compared with bupivacaine. Common concentrations of local anesthetic for peripheral nerve catheter patient-controlled analgesia (PNC PCA) include ropivacaine, 0.2%–0.3%, and bupivacaine, 0.12%–0.25% ( Table 13.3 ). Including opioids in PNC PCA solutions is probably unnecessary, as peripheral opioids may increase side effects without improving analgesia. The addition of clonidine to ropivacaine for PNC PCA does not improve analgesia.
|Catheter||Surgery||Patient-Controlled Analgesia Solution||Basal Rate (mL/h)||Demand Dose (mL)||Lockout (min)|
|Interscalene, infraclavicular||Rotator cuff repair; hand surgery||Ropivacaine 0.2%||6–8||2–4||20|
|Subgluteus or popliteal sciatic||Foot and ankle surgery||Ropivacaine 0.2%||5–8||3–5||20–60|
A continuous infusion of local anesthetics is generally used in PNC PCA, as this improves analgesia compared with bolus dosing only. A low dose continuous infusion (combined with a demand dose) reduces local anesthetic consumption without compromising analgesia, in comparison with continuous infusion alone. For moderately painful shoulder surgery, decreasing an interscalene catheter infusion of ropivacaine 0.2% from a basal rate from 8 to 4 mL/h provided similar analgesia, but the reduction caused a higher incidence of breakthrough pain incidence and sleep disturbance.
In addition to management of adult postoperative pain, PCA can also be used for the management of labor pain, pediatric postoperative pain, and cancer pain.
The most common modality for analgesia in labor is epidural. PCEA is a highly effective way of providing safe and superior labor analgesia. A multicenter, randomized controlled trial found that IV PCA and PCEA had the same rates of cesarean delivery or instrumental vaginal delivery. However, patients receiving IV PCA were more likely to receive antiemetic therapy, had more sedation, and more neonates in this group required naloxone and active resuscitation (52% vs. 31%). Patients receiving PCEA had better pain relief and greater satisfaction. The optimal choice of epidural infusion mixture and PCEA regimen is still debated. A meta-analysis comparing continuous epidural infusion (CEI) to PCEA without background infusion concluded that patients in labor receiving PCEA are less likely to require anesthetic interventions, require lower doses of local anesthetic, and have less motor block than those who received CEI. The addition of a basal rate to PCEA may further improve labor analgesia. Demand-only PCEA was associated with higher incidence of breakthrough pain, higher pain scores, shorter duration of effective analgesia, and lower maternal satisfaction, compared with PCEA with background infusion.
Despite the superiority of PCEA for pain control in labor, some parturients do not want epidural analgesia or have clinical conditions that contraindicate its use. In this situation IV PCA should be considered. Administration of opioids to parturients can cause newborn sedation or impaired respiration. Some practitioners limit exposure of the fetus to opioids by discontinuing IV PCA once the mother’s cervix is dilated. Compared with bolus parenteral opioids, IV PCA facilitates titration of analgesia as labor progresses, and can better compensate for interpatient variability in analgesic requirements. For this reason, IV PCA (compared with intermittent IM dosing) may provide better pain relief, and reduce maternal sedation, respiratory depression, and nausea. Compared with IM dosing, IV PCA for labor analgesia reduces umbilical cord blood opioid levels (indicating less placental drug transfer); in most cases IV PCA does not cause significant fetal depression. Use of shorter-acting opioids for labor IV PCA (such as fentanyl, alfentanil, and remifentanil) may reduce neonatal respiratory depression.
Pain Control in Pediatric Patients
PCA can reduce pain effectively and safely for adolescents and children. The critical determinant of successful PCA implementation in the pediatric population is the ability of the patient to understand the basic principles of PCA use. As a result, children younger than 4 years of age are not good candidates for PCA use. Children aged 4–6 years can use PCA pumps with the encouragement of nursing staff and parents. Nonetheless, the success rate in this age group is low. Children older than 7 years of age often can use PCA independently. Parental assistance for PCA use by young children has been advocated by some investigators. Implementation of formal parent education programs along with close observation by nursing staff is necessary if parent-controlled analgesia is to be considered. Parent-controlled analgesia bypasses the basic safety system of PCA and has been discouraged in the postoperative setting. Basal opioid infusions have also been successfully used by some physicians in the pediatric population for postoperative analgesia. However, some studies have shown an increased risk of hypoxemia in children receiving basal narcotic infusions with PCA. In a clinical context where continuous opioid infusions are deemed necessary, methods of detecting opioid-induced respiratory depression, such as pulse-oximetry, should be considered. In addition to the caution required for the use of continuous infusion in the pediatric population, concurrent administration of drugs with respiratory depressant effects should also be viewed with extreme vigilance. Typical PCA dosing for children via IV, epidural, and PNC routes are shown in Tables 13.4–13.6 .
|Drug||Bolus (μg/kg)||Lockout (min)|
|Drug||Basal Rate (mL.kg.h −1 )||Demand Dose (mL.kg −1 )||Lockout (min)||One-Hour Limit (mL.kg.h −1 )|
|Bupivacaine 0.06% + Hydromorphone 10 μg mL −1||0.1–0.3||0.1||Minimum of 10 min||Max = 0.4|
|Drug||Basal Rate (mL.kg.h −1 )||One-Hour Limit (mL.kg.h −1 )|
|Ropivacaine 0.2%||0.1–0.2||1.25 |
Note: the maximum safe dose is equivalent to 2.5 mg.kg.h − 1
Pain Control in Cancer Patients
PCA is one of the multimodal methods of effective cancer pain management in the inpatient setting for both adults and pediatric patients. The dosages of narcotics used in treating cancer pain often surpass those used postoperatively. Consequently, the utilization of basal continuous opioid infusions for the management of cancer pain is very valuable, and in contrast to postoperative pain management, should be encouraged. Parenteral narcotics provide a vital route for providing analgesia in patients with moderate to severe cancer pain. One study demonstrated changing the route of opioid administration, including the use of PCA-administered parenteral narcotics, is an important tactic for patients that have refractory cancer pain. Moreover, the use of methadone in PCA pumps, a practice uncommonly advocated for postoperative pain, is also a useful consideration in treating intractable cancer pain.
Pain Control in Patients With Obstructive Sleep Apnea
Strategies for pain management in patients with obstructive sleep apnea (OSA) emphasize the use of multimodal analgesia, long-acting local anesthetic infiltration of the surgical wound, regional techniques, and minimizing the use of sedatives and opioids. Several studies have demonstrated that opioids exacerbate both central and obstructive events in patients with OSA, placing these patients at higher risk of respiratory depression and respiratory failure. This increased sensitivity was also seen in patients on chronic opioid therapy (6 months or longer).
Current recommendations are that IV PCA should be used with caution in patients with known or suspected OSA. If an IV PCA is used, a basal infusion is not recommended, because the use of a basal rate is associated with increased risk of respiratory depression.
Opiate-containing EPCA is likewise associated with respiratory depression in patients with OSA. A recent meta-analysis including 121 patients suggested an incidence of major cardiorespiratory complications (including three deaths, one cardiorespiratory arrest, and two episodes of severe respiratory depression) of 4.1%. Significantly, five of these major complications occurred in patients receiving continuous fentanyl-based epidural infusions. The authors note that, to date, all studies are small and comprise low-quality evidence, and that the true rate of cardiorespiratory complications in OSA patients receiving neuraxial opiods is difficult to determine from the available data.
Patients with OSA on postoperative opioid PCA therapy should be monitored by continuous pulse oximetry. The appropriate setting for monitoring remains controversial: Although some evidence supports a step down or ICU location, the most recent ASA practice guidelines concluded that the literature is insufficient to offer guidance regarding either the setting or duration of monitoring. Finally, patients should have supplemental oxygen provided for hypoxic events, and home continuous positive airway pressure (CPAP) therapy should be continued for the entire hospital stay.