Introduction
Many survivors of critical illness face severe long-term health problems and QOL impairment [1]. These numerous disorders are described as post-intensive care syndrome (PIT syndrome) [2]. There are data on chronic pain in about 50% of PITS patients [3, 4]. Very little is known about predictors of post-intensive care chronic pain [5, 6]. As a rule, acute nociceptive pain in intensive care units is provoked by numerous factors. The most important ones are painful medical procedures and prolonged immobilization [7, 8]. Intraoperative nerve damage can result neuropathic pain [9]. Inflammatory processes may be followed by decrease of pain threshold and nociceptor tolerance [10—12]. There is more and more evidence that somatosensory function changes associated with pathology of small A-delta and C fibers responsible for sensation of cold, warmth and pain underlie post-intensive care [13—16]. These disorders are clinically manifested by pain syndrome and do not fall under classical definition of post-intensive care sensorimotor neuropathy. In recent studies, the authors found reduction of intraepidermal nerve fiber density in survivors of critical illness using skin biopsy as a standard diagnostic tool for verifying small fiber neuropathy [13—15]. These data confirm an assumption of severe damage to A-delta and C-fibers in these patients.
Non-invasive quantitative sensory testing (QST) has been used in the last decade to assess somatosensory function. QST can detect enhanced sensory function including reduced sensory thresholds (hyperesthesia and hyperalgesia) and loss of sensory function (hypoesthesia and hypoalgesia) [17, 18]. Changes in temperature sensitivity thresholds were associated with abnormalities in small myelinated A-delta fibers and unmyelinated C-fibers. Considering above-mentioned data, the main purpose of our study was to assess thermal QST parameters in survivors of critical illness, as well as to analyze prediction of chronic pain syndrome in these patients.
Material and methods
Patients
As soon as the local ethical committee has approved the study, we have been recruiting patients between 2015 and 2018. Patients were selected at the intensive care unit No. 1 of the Baranov Republican Hospital. Inclusion criteria were age 18—75 years, minimum ICU stay > 3 days, mechanical ventilation > 48 hours, APACHE II score ≥ 7 at admission, signed an informed consent. Exclusion criteria were previous stroke, traumatic brain and spinal cord injuries, thermal trauma, cancer, cardiac surgery and mental disorders. At discharge, patients were tested using the pain questionnaires. They also underwent thermal quantitative sensory testing. The second stage of examination was carried out 6 months later. We analyzed follow-up data, performed neuro-orthopedic examination and repeated testing using pain questionnaires.
Questionnaires
Localization and intensity of pain were assessed using the short-form pain questionnaire. Neuropathic pain was estimated using the PainDetect questionnaire (<12 scores – no neuropathic pain, 12—18 scores – unclear result, > 18 – neuropathic pain).
Neuro-orthopedic examination
We analyzed impairment of all types of sensitivity in key sensory points. Active movements were sequentially assessed in all joints with analysis of pain intensity.
Diagnosis of post-intensive care chronic pain
Chronic post-intensive care pain in 6 months after discharge was defined as clinically significant pain de novo lasted at least 3—6 months and associated with ICU stay. Patients were asked to indicate whether they experience any persistent pain after discharge from ICU (only pain de novo that they did not have before admission to ICU). In case of similar pain, they were asked to describe the affected body parts graphically as in the short-form pain and PainDetect questionnaires.
Clinical data
Anamnestic data were collected from the outpatient records. We applied the Charlson comorbidity index (CCI) to assess the baseline comorbidities. Any surgeries or traumas, that could potentially cause chronic pain (any drainage tubes, especially pleural drainage) were also considered. Duration of mechanical ventilation was defined as the number of days with invasive respiratory support. Length of ICU-stay was calculated in complete days. Length of hospital-stay was defined as the total number of days between primary admission and discharge. We used APACHE-II and SOFA scores to determine severity of critical illness at the ICU. Maximum serum C-reactive protein (mg/L) was considered as a marker of inflammation.
Quantitative sensory testing
Each patient was instructed on the protocol for quantitative sensory testing. We used the TSA-II neurosensory analyzer with a standard thermode 30 × 9 × 30 mm (Medoc, Israel). Cold detection thresholds (CDT), warm detection thresholds (WDT), paradoxical heat sensation (PHS), cold pain thresholds (CPT) and heat pain thresholds (HPT) were analyzed. Thermal thresholds and pain were evaluated within key sensory points L5 on the lower limb and C6 on the upper limb.
Statistical analysis
Statistical analysis included descriptive statistics (central tendency parameters and variance of quantitative variables, as well as absolute values and percentages for categorical and binary values). Between-group comparison of means was performed using the Student's t-test. In case of abnormal distribution of sample, was used non-parametric methods (Mann – Whitney or Kruskal – Wallis tests) for between-group analysis. Statistical significance of between-group differences of frequencies was tested using the χ2 test. Differences were considered significant at p-value < 0.05. One of the unsupervised learning method (hierarchical cluster analysis) was used to identify the groups of critically ill patients with similar QST parameters and compare these groups later. Input data for HCA consisted of left- and right-sided QST cold and warm thresholds measured in dermatomes L5 and C6. The Euclidean distance was taken as a metric of between-patient distance. Cluster analysis was carried out using the WARD method.
Results
The final sample included 99 patients with surgical diseases who were treated at the ICU and subsequently discharged. Patient distribution by gender and age is shown in Table 1.
Table 1. Survivors of critical illness distribution depending on gender and age
|
Age, years |
Gender |
Total |
Female, % |
|
|
Male |
Female |
|||
|
<45 |
16 |
22 |
38 |
57.9 |
|
<60 |
9 |
25 |
34 |
73.5 |
|
<75 |
13 |
13 |
26 |
50.0 |
|
≥75 |
0 |
1 |
1 |
— |
|
Total |
38 |
61 |
99 |
61.6 |
Women prevailed in the sample. They made up to 75% among patients aged 45 — 59 years.
Re-evaluation after 6 months revealed chronic pain syndrome in 58 (59%) patients. The main characteristics of patients with and without pain after 6 months are summarized in Table 2.
Table 2. Characteristics of patients with and without chronic pain syndrome after 6 months
|
Variable |
Value (n=99) |
Patients with chronic pain (n=59) |
Patients without chronic pain (n=40) |
p |
||
|
Mean |
Me (Q1; Q3) |
Mean |
Me (Q1; Q3) |
|||
|
Age, years |
49±14.2 |
51.0±14.8 |
51.0 (39.0; 64.0) |
46.5±13.1 |
47.0 (34.0; 57.0) |
0.113 |
|
Ventilation ,days |
16.6±18.9 |
21.9±21.9 |
16.0 (7.0; 32.0) |
8.8±9.6 |
5.0 (3.0; 10.0) |
<0.0001 |
|
CRP, mg/L |
220.9±102.2 |
241.1±102.3 |
223.0 (182.0; 320.0) |
191.2±95.8 |
198.5 (121.8.0; 232.0) |
0.026 |
|
ICU-stay, days |
23.6±22.1 |
30.2±25.4 |
22.0 (12.0; 44.0) |
13.6±9.6 |
10.0 (7.0; 16.0) |
<0.0001 |
|
Hospital-stay, days |
47.3±30.3 |
55.4±32.2 |
49.0 (32.0; 68.0) |
35.4±22.9 |
29.0 (19.0; 42.0) |
<0.0001 |
|
CCI score |
2.0±2.1 |
2.2±2.3 |
2.0 (0.0; 4.0) |
1.8±2.0 |
1.5 (0.0; 3.0) |
0.406 |
|
APACHE II score |
16.6±5.0 |
16.9±4.3 |
16.0 (15.0; 19.0) |
16.2±5.9 |
15 (11.5; 18.5) |
0.162 |
|
SOFA score |
8.4±2.5 |
8.8±2.4 |
9.0 (8.0; 10.0) |
7.8±2.5 |
8.0 (6.0; 10.0) |
0.035 |
|
CCI >0 |
48 (48.5%) |
36 |
61.0% |
22 |
55.00% |
0.698 |
|
Drainage tubes |
||||||
|
No |
6 (6.0%) |
2 |
3.40% |
4 |
10.00% |
0.293 |
|
1 tube |
92 (92.9%) |
56 |
94.90% |
36 |
90.00% |
|
|
2 tubes |
1 (1.0%) |
1 |
1.70% |
0 |
0.00% |
|
|
Pleural drainage tubes |
||||||
|
No |
44 (44.4%) |
20 |
33.90% |
24 |
60.00% |
0.036 |
|
1 tube |
48 (48.5%) |
34 |
57.60% |
14 |
35.50% |
|
|
2 tubes |
7 (7.0%) |
5 |
8.50% |
2 |
5.00% |
|
Note. CCI — Charlson comorbidity index; CRP — C-reactive protein; ICU — intensive care unit.
We found significant differences in cold and heat perception thresholds between patients with and without chronic pain (Fig. 1).
Fig. 1. Heat and cold detection thresholds in patients with and without chronic pain.
For an in-depth study of QST changes and assessment of the influence of various factors, we carried out a hierarchical cluster analysis of the entire sample using standardized values of temperature thresholds. Standardization was carried out relative to mean and standard deviation of each parameter of heat and cold thresholds for dermatomes L5 and C6 on both sides.
The final hierarchical tree demonstrates a clear division of patients into 4 groups. Volumes of these groups still allow statistical analysis (Fig. 2). We found between-cluster differences for all thermal thresholds (Fig. 3).
Fig. 2. Distribution of patients by four clusters.
Fig. 3. Comparison of distribution of QST thresholds for cold and heat.
Normal thermal thresholds were noted only in cluster 1. In the 2nd and 4th clusters, increased thermal thresholds demonstrated thermal and cold hypoalgesia (“deafferentation nociceptor”). Thresholds were reduced in the 3rd cluster (“irritated nociceptor”). The concept of “irritated nociceptor” is based on a sensory phenotype with intact function of small fibers (cold and heat sensation) and hyperalgesia. At the same time, reduced (or even lost) perception of heat and cold dominated in the “deafferentation nociceptor” profile [3, 19].
Between-cluster analysis of mechanical ventilation revealed significant correlation of respiratory support time and QST parameters in critically ill patients (Fig. 4a).
We found minimal time of ventilation in the 1st cluster (<5 days in most patients and < 10 days in all cases). In the 3rd cluster, duration of mechanical ventilation was 5 — 15 days as a rule. In the 2nd cluster, all patients required ventilation for at least 20 days (20 — 30 days in most cases). In the 4th cluster, all patients except for 2 ones required respiratory support for more than 30 days. Clear division into QST thresholds led to an equally clear division into duration of mechanical ventilation, although this parameter was not involved in cluster analysis. Comparing these results with previous data on QST, we can assume normal QST parameters in case of mechanical ventilation within a week. Prolonged ventilation is associated with more severe changes in QST thresholds. We found significant between-cluster differences in SOFA score in all 4 clusters. However, CRP levels were similar (Fig. 4b, c). There were no significant between-cluster differences in APACHE II scores due to high variability of these values. Nevertheless, typical trends are clearly observed on scatter plots (Fig. 4d).
Fig. 4. Comparison of duration of mechanical ventilation (a), C-reactive protein (b), SOFA (c) and APACHE II (d) scores between patients in four clusters.
We analyzed pain intensity using verbal rating scale between all clusters and found a correlation between pain syndrome and QST patterns for dermatomes L5 and C6 (Fig. 5a). Indeed, pain was significantly more intense in clusters 2 and 4 compared to cluster 3 and especially cluster 1. However, pain intensity was similar in clusters 2 and 4. There was significant difference in severe pain score between clusters 1 and 3.
Comparison of clusters with different sensory profiles of QST showed that patients with a profile of deafferentation nociceptor had more severe pain and higher PainDetect score that indicated neuropathic pain syndrome (Fig. 5b). Between-cluster analysis of PainDetect data also demonstrate the relationship between PainDetect and QST patterns (Fig. 5b). In all 4 clusters, patients clearly differed regarding PainDetect score. In the 1st cluster, almost all patients had PainDetect score < 12. In the 3rd cluster, 75% of patients rated pain intensity over 11 scores. It was found that 75% of patients in cluster 2 and all patients in cluster 4 had PainDetect score > 20. Moreover, 75% of patients in the 4th cluster rated pain intensity ≥ 24—25 scores.
Fig. 5. Distribution of NRS (a) and PainDetect (b) scores.
Between-cluster comparison of pain sensory phenotypes according to the PainDetect questionnaire was interesting (Fig. 6).
Fig. 6. Severity of spontaneous and provoked pain between various clusters.
Between-cluster comparison of pain profiles and QST patterns ensures dynamic assessment of pain phenotypes depending on duration of mechanical ventilation and severity of critical illness. Moreover, we can establish the association between pain profiles and QST patterns. For example, spontaneous (burning) and provoked pain (exposure to temperature) was the most severe in the 3rd cluster that indicated formation of an irritated nociceptor. In case of prolonged ventilation and more severe critical state, sensory phenotype indicated a deafferentation nociceptor. Numbness and allodynia were more obvious in these clusters (2 and 4).
Discussion
Chronic pain is increasingly recognized as a problem in post-intensive care survivors [18]. There are contradictory data on the incidence of chronic pain. According to our study, over 50% of post-intensive care survivors experienced pain within 6 months after discharge. Foreign colleagues reported similar data on chronic pain in 56% of patients within 2 years after discharge from intensive care unit [5, 19]. Various authors found reduction of the incidence of chronic postoperative pain over time, but this statement does not apply to chronic post-intensive care pain.
We found no significant correlation between chronic post-intensive care pain and age. At the same time, Choinière M. et al. [6] and Baumbach P. et al. [20] reported higher risk of chronic post-intensive care pain in younger people. On the contrary, C.E. Battle et al. (2013) found that older age is associated with chronic post-intensive care pain while other studies did not confirm these data [5, 21]. Similar differences may be potentially explained by features of samples. Age-related increase in the risk of chronic post-intensive care pain is possibly associated with normal physiological changes, such as muscle involution, decrease of metabolic rate, often with nutritional deficiency, and age-related changes in the musculoskeletal system [22].
In experimental conditions, women have a higher risk of chronic pain and higher pain sensitivity [23]. The underlying mechanisms (for example, endocrine or psychosocial differences) are still being studied, and there is no definite answer yet. Battle C.E. et al [5] found no correlation between gender and chronic post-intensive care pain. Scientific data on relationship between chronic postoperative pain and gender are also controversial [21, 24].
High CRP at the intensive care unit is associated with increased risk of chronic post-intensive care pain. These data emphasize potential role of inflammatory processes in chronic pain [20]. Comparison of patients with and without chronic pain revealed significantly higher CRP in the 1st group. Cluster analysis found no significant differences in CRP levels. Therefore, we can suppose no clear relationship between QSR parameters and CRP as a marker of inflammatory process. Battle C.E. et al. [5] found sepsis as a significant independent predictor of pain in 6 months after discharge from ICU. Perhaps, certain combination of sepsis and CRP level may be essential in more frequent development of chronic pain [25].
Duration of mechanical ventilation and length of ICU stay are significant risk factors of chronic pain. Battle C.E. et al. [5] confirmed this fact in univariate analysis. At the same time, T.K. Timmers et al. [25] did not determine length of ICU stay and duration of mechanical ventilation as significant risk factors of chronic pain persisting after discharge. Perhaps, this is due to various diseases in patients. For example, Timmers T.K. et al. analyzed patients of therapeutic profile.
Baumbach P. et al. [26] confirmed a correlation between dysfunction of small fibers, pain and delayed recovery of physical activity. According to our data, thermal QST after discharge from ICU can have additional prognostic value for identifying patients with high risk of severe chronic pain syndrome. These patients can benefit from early rehabilitation measures preventing chronic pain [27].
Conclusion
It was revealed that 59% of post-intensive care patients suffer from pain within 6 months after discharge. Risk factors of chronic post-intensive care pain are duration of mechanical ventilation, high SOFA score and high CRP level. Thermal QST is valuable to identify patients with small fiber dysfunction and analyze post-intensive care changes. It is important for multidisciplinary approach to pain relief in these patients.
Author contribution
All authors participated in data analysis, writing and editing the manuscript, finally approved the version to be published and agree to be responsible for all aspects of the study.
The authors declare no conflicts of interest.