Chronic pain is increasingly recognized as a problem in survivors of critical illness. Despite an increased awareness of its contribution to reduced quality of life, pain remains a significant problem for survivors of critical illness [5]. Conflicting evidence exists regarding the incidence of chronic pain in this patient group. The results of this study, however, highlight that nearly half of all respondents still experience pain at least 6 months after ICU discharge. This concurs with a previous study that reported that 56% of patients still reported pain at 2 years after ICU discharge [3]. In a Dutch study by Hofhuis et al., 2008 [19], health- related quality of life was reported to be significantly lower in critically ill patients at pre-ICU admission when compared with the healthy Dutch population. They also reported that health-related quality of life remained significantly lower than that of the healthy Dutch population at 6 months after ICU discharge, except in the bodily pain score, which does not concur with the results of this study [19].
This study reported that more than 20% of patients were experiencing shoulder pain at least 6 months after ICU discharge. This result is higher than that of an age-matched general population of 15% in a study in the Netherlands in 2011 and 11.7% in a British study in 2002 [20, 21]. In a survey of chronic pain in Europe and Israel, chronic shoulder pain was reported in 9% of the population [22]. In a recent small study completed in England in 2012, 80% of ICU survivors were reported to have shoulder dysfunction over a period of 1 year after ICU discharge [23]. Limited previous research exists investigating specific body parts affected by pain in survivors of critical illness. A possible contributory factor for the incidence of shoulder pain may include the patient's reluctance or inability to move the shoulder girdle because of the position of the central line, dialysis lines, and ventilator tubing. Another possible cause of shoulder pain is the lack of muscle tone during critical illness. The shoulder joint is well recognized as an unstable joint when lacking muscle tone [24], and this may lead to chronic pain in survivors of critical illness. One reason for this is that the shoulder is potentially put under undue strain during frequently performed nursing procedures, such as rolling, at a time when it is at its most vulnerable. An increased awareness of the incidence of shoulder pain in patients discharged from ICU should encourage healthcare professionals responsible for caring for these patients to take appropriate measures to handle the shoulder joint appropriately at all times.
Nearly one third of all patients reported using healthcare resources in an attempt to address their chronic pain, which has potential cost implications for provision of ongoing care for these patients. In the survey by Breivik et al. (2006) [22], 60% of patients with chronic pain had visited their GP about their pain two to nine times in the last 6 months [22]. In a study by Der Schaaf et al. (2009) [11], it was reported that at 3 months after hospital discharge, 45% of all patients were following an interdisciplinary rehabilitation program to address their ongoing functional needs [11]. Recent emphasis has been placed on the need for follow-up of this patient group and the provision of rehabilitation programs (NICE 2009), and our study's reports supports these recommendations [25].
No significant differences were found between the patients with chronic pain and those without, in terms of gender, primary admission criteria, or APACHE II scores. Dowdy et al. (2005) [4] similarly concluded in their systematic review that gender and admission diagnosis were not predictors of quality of life. Dowdy et al. (2005) [4] did report, however, that a trauma diagnosis was a predictor of pain; however, because of the small number of trauma patients in this study, cross-comparison is not possible.
The significant risk factors for chronic pain on the univariate analysis were severe sepsis and ICU length of stay. Primary admission diagnosis of surgery, wound, ICU and hospital length of stay were dropped as significant risk factors in the multivariable analysis, and this may reflect the influence of confounding in an observational study of this design. In a study by Timmers et al. (2011) [9], ICU length of stay and mechanical ventilation days were also not reported to be significant risk factors for chronic pain after discharge from ICU, which concurs with the results of the multivariable analysis in this study.
On multivariate analysis, one of the significant risk factors for chronic pain was severe sepsis. These results concur with a German study by Zimmer et al. (2006) [7], in which patients who had survived severe sepsis were also reported to have significantly higher levels of pain, although the time since discharge is unclear. One potential explanation for this reported chronic pain was suggested by Zimmer et al. (2006) [7], who described the proinflammatory cytokine response that has been demonstrated to increase pain intensity. Another possible explanation is that the patient with severe sepsis often requires high levels of inotropic support, thus contraindicating early mobilization and rehabilitation. Further prospective studies are required to investigate possible mechanisms for the influence of severe sepsis on chronic pain.
The patient's age was another significant risk factor for chronic pain in patients discharged from ICU. This may be due to the normal physiologic changes associated with increased age, such as reduced muscle mass, reduced levels of blood and tissue metabolites, and a poor nutritional status. A number of studies have reported that age influences the patient's functional recovery, and these are summarized in the review by Dowdy et al. (2005) [4]. In contrast to the results of this study, however, Dowdy et al. (2005) reported that age did not influence pain experience at 6 months after ICU discharge [4].
Younger age has been reported as a risk factor in investigations of chronic pain after surgery; however, these results remain inconsistent [26]. A number of mechanisms for this have been proposed and relate primarily to reduction in peripheral nociceptive function with increased age [26]. In contrast to the postsurgery studies, increased age may be a risk factor for chronic pain in post-ICU patients because of a number of other potential mechanisms. These include the normal physiologic ageing processes affecting the musculoskeletal system, an increased number of comorbidities, and higher rates of polypharmacy evident in the elderly population. In this study, presence of a surgical wound was not a risk factor for chronic pain on multivariable analysis.
This study highlighted the incidence of chronic pain and its potential risk factors at 6 months to 1 year after ICU discharge. One interesting finding of this study was that the existence of preexisting chronic pain conditions before ICU admission was not a risk factor for chronic pain after discharge from the ICU. An increased awareness of the incidence of chronic pain should ultimately result in an attempt to preempt it, through the aggressive use of therapeutic interventions for pain. Early mobilisation and rehabilitation during the ICU stay is increasingly considered one of the most effective strategies for reducing pain and functional disability after discharge, but further research is still needed [27]. Simply having an understanding of the potential causes of ongoing shoulder pain and how to minimize the risk of its occurrence, such as appropriate handling of the joint, could improve long-term outcomes in this patient group.
The use of ICU follow-up clinics was recommended by the National Institute of Clinical Excellence (NICE) in 2009, and these clinics provide the opportunity to address ongoing pain and dysfunction [25]. Further prospective studies are needed investigating long-term outcomes in survivors of critical illness and possible therapeutic interventions to reduce chronic pain.
The present study has a number of limitations. As a result of the study design and the inherent nature of patients, a number of the risk factors investigated were potentially interdependent, so an increase in one variable inadvertently results in an increase in another. Multivariable logistic regression with backward elimination techniques was used to address this issue of collinearity. In clinical research, however, this is difficult to overcome because of the nature of the study population, and therefore, the results of this study should be interpreted with this in mind.
A 6- month cohort was used to include a sufficient number of patients to provide valuable results. It could be suggested that a patient's pain symptoms may potentially change over the 6- month time period used in this study; however, in a study of this nature, this is unavoidable. It is also questionable as to how significantly a patient's pain symptoms would change over a 6-month period, because of the nature of chronic pain. One method of overcoming this would be to complete a multicentered study over a shorter time period.
A sample size of 100 patients with chronic pain was needed for this study according to Peduzzi et al. (1995) [12], but as only six variables were entered into the final multivariable analysis, the sample size was considered sufficient for the final analysis. The low response rate (61%) may have introduced nonresponse bias; however, nonresponder analysis was used in an attempt to control for this bias. The use of interviews to complete the questionnaires and the methods used to handle missing data in this study may also have inadvertently led to a degree of interviewer or interpretation bias. As a result of these different types of potential bias, the results of this study should be interpreted with caution.
No suitable previously validated questionnaire was available for use in this study. Therefore, a further limitation of this study was the use of a newly designed questionnaire. The study results may lack validity and generalizability as a result. The questionnaire was piloted on a number of survivors of critical illness to help to overcome this limitation. Another limitation of investigating outcomes in survivors of critical illness is the high mortality rate and loss to follow-up. This may introduce a degree of reporting bias, but this is unavoidable in a study of this type.
In an observational study using a retrospective database analysis, the potential exists for information bias. Data may have been incorrectly entered into the database at the time of the patient's admission, or the data may be incorrectly copied for use in the study. A validation check was used in an attempt to overcome this error of data extraction and data input and thus reduce information bias. A further limitation is the potential underreporting of severe sepsis by doctors in ICU. The aim initially was also to include ARDS in our study, but because of the obvious underreporting of ARDS data, the decision was made to exclude ARDS as a risk factor from our study. Availability of further information regarding other factors important to pain experience (for example, medications used during the ICU stay) would have improved the reliability and validity of this study. A further prospective study would be needed to overcome these limitations, and this should be considered when interpreting the study results.