Criterion validity and discriminatory ability of the central sensitization inventory short form in individuals with inflammatory bowel diseases

Introduction

Inflammatory bowel diseases (IBD), including Crohn’s disease and ulcerative colitis, are chronic inflammatory conditions of the gastrointestinal tract [1], [2], [3], [4]. Pain is reported in over 70% of individuals with IBD, where the most common complaints relate to abdominal and musculoskeletal (MSK) pain experiences [5], [6], [7], [8], [9]. Although acute abdominal pain is a common consequence of active IBD and structural changes, such as strictures, current literature reports abdominal pain experiences in the absence of such factors, suggestive of mechanisms of central sensitization [10], [11]. Our previous investigations exploring persistent MSK pain in IBD demonstrated worse pain experiences in individuals with greater symptoms related to central sensitization [12], [13].

Central sensitization has been proposed as a phenomenon that involves hypersensitivity of the central nervous system [14], [15], [16], [17], which may lead to worse pain experiences seen in multiple patient populations, including low back pain, osteoarthritis, and rheumatoid arthritis [18], [19], [20]. The mechanisms of central sensitization are believed to be responsible for the overlapping clinical features of central sensitivity syndromes, including irritable bowel syndrome, fibromyalgia, and temporomandibular joint disorder [21], [22]. It has been proposed that the symptoms common to these syndromes, including fatigue, sleep disturbances, and cognitive symptoms, can be considered not as belonging to individual disorders, but as different manifestations of primary central sensitization mechanisms [22]. This viewpoint informed the development of symptomatic screening tools, such as the central sensitization inventory (CSI), to broadly assess and quantify the degree of presenting symptomatology, in order to identify potential contributions from mechanisms of central sensitization [22].

The original CSI is a 25-item tool used to investigate an array of common symptoms associated with central sensitivity syndromes, within the domains of physical, emotional distress, headache/jaw, and urological features [22]. Higher CSI-25 scores indicate increased symptomology, with a benchmark score (≥40) suggesting the likely presence of one or more central sensitivity syndromes [22], [23]. Additionally, total CSI-25 scores have demonstrated the ability to distinguish between patients with chronic MSK pain and healthy controls [23], [24]. Similarly, higher CSI-25 scores have demonstrated association with worse pain-related outcomes [24], [25], as well as poor long-term postoperative outcomes [26], [27].

We previously utilized CSI-25 as a screening tool for central sensitization symptomology in patients with IBD, where worse pain experiences were associated with higher CSI scores [12], [13]. Like many chronic conditions, pain management has been identified as a top challenge and priority in IBD research, indicated by both clinicians and patients [28], [29], [30]. As such, linking the presence of pain with a profile of symptomology related to central sensitization may provide clinicians with useful information regarding the potential presence of central mechanisms. Similarly, CSI may represent a useful tool for prospectively monitoring patients’ risk for developing persistent pain, as well as responses to treatment.

Although CSI-25 has previously been used in IBD research, a potential limitation relates to the inclusion of items exploring symptoms commonly seen in IBD patients consequent of underlying disease processes and inflammation, such as bowel and skin problems [31]. CSI-25 does not differentiate whether positive scoring of these items is related to symptoms of central sensitization or are a consequence of active IBD. Recently, a 9-item short form of the CSI (CSI-9) was developed as a quick and clinically useful screening tool to reduce patient burden in both clinical and research environments [32]. Interestingly, the items which potentially overlap between common symptoms of IBD and those of central sensitization have been excluded in the CSI-9, suggesting that CSI-9 may be more useful to explore central sensitization symptoms in IBD patients.

As previously noted, the CSI-25 benchmark score (≥40) was validated to identify the presence of central sensitivity syndromes [22], [23], and previously demonstrated correlation to chronic MSK pain and worse pain experiences [33]. However, targeted investigations in specialized populations, such as migraine [34], fibromyalgia, and other MSK pain experiences [35], have demonstrated unique CSI-25 cut-off scores to discriminate conditions and/or pain presentations in these populations. Similarly, IBD-specific cut-off values for CSI could provide clinicians with a tool which highlight symptoms that may persist and contribute to possible central sensitization mechanisms alongside pain presentations, thereby triggering targeted management pathways. Therefore, investigation of appropriate cut-off values for both CSI-25 and CSI-9 are needed to better understand the use of CSI tools in IBD patients.

The aim of the present study was to evaluate the performance of CSI-9 compared to the original CSI-25 in individuals with IBD. Study objectives were to investigate the criterion validity of the CSI-9 to the CSI-25, assess individual association of the CSI measures with clinical features of IBD and pain presentations, and to establish disease-specific CSI-9 and CSI-25 cut-off scores for discriminating the presence of self-reported pain (i.e. abdominal and MSK) in individuals with IBD. We hypothesized that CSI-9 would demonstrate acceptable validity with similar association to participant features as the CSI-25.

Methods

Results

A total of 370 individuals with IBD volunteered to participate in the online survey. Eleven respondents were excluded due to minimum age requirements. Thirty-nine respondents were excluded due to incomplete essential datasets. The mean (SD) age of the remaining 320 respondents was 43.86 (14.76) years, with 80% (n=256) of participants identifying as female, 19% (n=62) as male, and 1% (n=2) as gender diverse. Study participants identified with the following ethnic groups: New Zealand European (n=286), Maori (n=19), Indian (n=5), Australian (n=5), North American (n=4), Fijian (n=3), South African (n=2), and other European (n=14).

Discriminate analysis

ROC analysis (Figure 1) and AUC (95% CI) indicated that CSI-9 (0.788 (0.725, 0.851), p<0.001) and CSI-25 (0.808 (0.750, 0.867), p<0.001) were able to adequately discriminate between participants with and without pain. The maximum Youden Index suggests that the optimal cut-off score to discriminate individuals with self-reported abdominal and/or MSK pain is ≥17 for CSI-9 and ≥40 for CSI-25 (Table 3). A cut-off value of 17 for CSI-9 represented a sensitivity and specificity of 83.5 and 61.7%, respectively, whereas a cut-off of 40 for the CSI-25 demonstrated a sensitivity and specificity of 73.5 and 76.7%, respectively.

Figure 1:

Receiver operating characteristic curve analyses for the 9-item central sensitization inventory (CSI-9) and the original 25-item CSI (CSI-25) to discriminate the presence of self-reported pain in individuals with IBD.

Table 3:

Central sensitization inventory original and short form cut-off scores for discriminating the presence of self-reported pain experiences in patients with IBD.

Version	Cut-off score	Sn, %a	Sp, %b	YI	AUC (95% CI)
CSI-9	15	88.8	48.3	0.372	0.788 (0.725, 0.851)*
	16	86.5	56.7	0.432
	17	83.5	61.7	0.451
	18	78.5	65.0	0.435
	19	71.9	70.0	0.419
	20	66.2	73.3	0.395
CSI-25	37	77.7	61.7	0.394	0.808 (0.750, 0.867)*
	38	75.8	68.3	0.441
	39	75.0	73.3	0.483
	40	73.5	76.7	0.501
	41	70.4	78.3	0.487
	42	67.7	80.0	0.477

1. Bold font represents cut-off values indicated by the highest Youden’s Index. IBD, Inflammatory bowel disease; Sn, sensitivity; Sp, Specificity; YI, Youden’s Index; AUC, area under the curve; CI, confidence interval; CSI, central sensitization inventory. *Significant at p<0.001. ^aSensitivity: the proportion of actual positive results that were correctly identified. ^bSpecificity: the proportion of negative results that were correctly identified.

Discussion

The purpose of this study was to evaluate the performance of the CSI-9 compared to the original CSI-25 in individuals with IBD. This study found a substantial agreement between the CSI measures (ICC_3,1=0.64) in this population, with similar associations to multiple participant features and unique cut-off values for discriminating the presence of self-reported pain. These results are similar to previous investigation of the psychometric properties of the CSI-9 in individuals with MSK pain (n=505), where the ICC_3,1 between CSI-25 and CSI-9 was 0.69 (95% CI, 0.64–0.73) [32].

Similarly, study results support previous reports which have demonstrated that associations with clinical features did not differ significantly between the two versions of CSI [32], [35]. In the present study, associations between CSI-25 and CSI-9 to SIBDQ (rho=−0.65 and −0.58, p<0.001) and abdominal pain severity (rho=0.27 and 0.19, respectively, p<0.001) represent the greatest differences between the CSI tools in their strength of association with clinical features. SIBDQ assesses numerous consequences of IBD on an individual’s quality of life, including bowel and systemic symptoms [45]. As previously indicated, some of the items omitted from the CSI-9 also examine symptoms commonly seen in IBD patients as a consequence of disease processes [31]. As such, the removal of these items may explain the weaker association between the CSI-9 and SIBDQ scores compared to the original CSI-25.

Current study results suggest that abdominal pain severity in patients with IBD is poorly associated with both CSI tools. Conversely, MSK pain experiences (i.e., severity and interference) demonstrated moderate associations with the CSI tools. This difference between pain severity correlations suggests that abdominal pain severity may be less related to central sensitization symptomology, and thus likely to be related to other constructs (i.e., structural disease processes). However, it is important to note that associations in the present study do not account for influences from factors which have previously demonstrated independent association with CSI and pain severity scores, such as sex [32], [41], comorbidity count [41], and IBD activity scores [13]. As such, it is unclear whether accounting for these additional factors would result in stronger observed associations between abdominal pain severity and CSI scoring.

In the present study, AUC results suggest that CSI-9 and CSI-25 are able to adequately discriminate the presence of self-reported pain in individuals with IBD, using cut-off values of 17 and 40, respectively. However, although the sensitivity and specificity of the CSI-25 cut-off point were relatively similar (i.e., 73.5 and 76.7%, respectively), the cut-off for CSI-9 demonstrated a greater difference (i.e., 83.5 and 61.7%, respectively). This suggests that although CSI-9 was able to more accurately identify participants with self-reported pain compared to the CSI-25, it was less accurate in correctly identifying the absence of pain. As such, although CSI-9 may be useful as an initial screening tool to identify IBD patients with an increased likelihood of having pain, positive findings should prompt further confirmatory investigations.

Current results demonstrate significant differences between high and low symptom groups for both CSI versions with regards to correlations with multiple participant features. Generally speaking, participants in the high CSI groups were more likely to be female and found to have worse health-related quality of life, greater comorbidity, and worse pain experiences. Although there were more female participants (n=256, 80%) in the present study, these results are similar to previous reports in multiple painful MSK conditions, suggesting that CSI may be capturing a similar construct in individuals with IBD [22, 23, 55, 56]. However, future research should utilize gender matched investigations to confirm study findings. Finally, with the exception of abdominal pain severity, all study features trended in a very similar manner across both CSI versions. This suggest that the items omitted in the CSI-9 may have little if any impact in determining the profile of the high/low symptom groups in individuals with IBD.

As previously stated, abdominal pain is considered a primary symptom of active IBD processes. Therefore, it is unsurprising that the percentage of participants with abdominal pain across the groups were very similar to those who met criteria for active IBD, with a similar trend seen in both CSI versions (Table 4). Additionally, although abdominal pain severity was not found to be significantly different between the CSI-9 groups, the interpretation of the severity values seen in the CSI-25 groups vs. those of CSI-9 are unlikely to be clinically meaningful.

This is the first study to explore and validate the use of CSI-9 in individuals with IBD. Study findings report notable similarities between associations of multiple participant features and both CSI tools. Consequently, study results suggests that the CSI-9 is suitable to use as a brief tool in patients with IBD, with suggested cut-off value of 17 to screen for individuals with pain (i.e., abdominal and/or MSK). However, the use of CSI-9 to monitor patients’ response to treatments and/or the development of chronic pain is still unknown. Therefore, future research should explore the use of continuous CSI-9 scoring as a clinical tool to prospectively identify patients at-risk of developing chronic pain and/or to monitor treatment responses. Additionally, future research exploring the relationship between other IBD clinical features and both CSI versions may provide greater depth of understanding of current study findings and for the use of these measures in this population. The present study included an online survey precluding the ability to utilize clinical assessments in study participants, which may have strengthened study findings. CSI-9 in the present study was extracted from one administration of the original CSI-25. Future research investigating independent administration of CSI-9 would allow for more definitive conclusions regarding its performance in individuals with IBD when compared to the CSI-25.