Abstract
Objective:
Pain is common in inpatient rehabilitation patients; however, the prevalence of pain diagnoses in this population is not well-defined. This study examines comorbid pain diagnoses in inpatient rehabilitation patients across impairment groups.
Design:
Adult inpatient rehabilitation patients discharged from January 2016 through December 2019 were identified in the Uniform Data System for Medical Rehabilitation database using a literature-established framework containing International Classification of Diseases, Tenth Revision, Clinical (ICD-10-CM) pain diagnoses. Demographic data, clinical data, and pain diagnoses were compared across the 17 rehabilitation impairment groups.
Results:
Of 1,925,002 patients identified, 1,347,239 (70.0%) had at least one International Classification of Diseases, Tenth Revision (ICD-10) pain diagnosis. Over half of all patients in each impairment group had at least one pain diagnosis. The most common pain diagnoses were limb/extremity and joint pain, with variation between impairment groups. Female sex and being in the arthritis, major multiple trauma, and pain syndrome impairment groups were associated with a greater odds of a pain diagnosis.
Conclusions:
Over half of all patients in each rehabilitation impairment group have a pain diagnosis, which varies between impairment groups. Because of the high prevalence of pain diagnoses, a new focus on pain management in inpatient rehabilitation patients is needed. Rehabilitation outcomes may also be affected by pain.
Keywords: Pain, Analgesia, Inpatient, Rehabilitation
Pain is a common diagnosis in inpatient rehabilitation units that is frequently managed by physiatrists using physical modalities, referrals to physical and occupational therapies, and analgesics. Despite the renewed focus on pain and emphasis on nonopioid interventions due to the ongoing opioid epidemic, the prevalence of pain in hospitalized patients remains high with one study suggesting up to 84% of patients affected.1 While frequently encountered in the postacute care rehabilitation setting, the prevalence of pain and corresponding diagnoses vary in the literature due to heterogeneity between studies and clinical differences between patients with different indications for rehabilitation. For example, of patients undergoing rehabilitation for a spinal cord injury, 97% reported having pain at least once during rehabilitation.2 Conversely, of patients undergoing neurorehabilitation in Germany, 25% had pain at the beginning of rehabilitation and the total number of patients with clinically relevant pain increased during rehabilitation in this study.3 In terms of pain severity, assessment of rehabilitation patients with traumatic brain, spinal cord, and burn injuries showed significantly higher pain intensities in these patients compared to the general population.4
While the prevalence of pain in inpatient rehabilitation patients and types of pain experienced by these patients is less clear, there is evidence that pain in inpatient rehabilitation patients impacts therapeutic interventions and rehabilitation outcomes, suggesting a strong need for prioritized management. For example, in patients with spinal cord injuries, those with severe pain (28.4%) demonstrate fewer hours with physical therapy and the need for modification of therapy sessions to focus on pain management.2 Similarly, increased pain is associated with a longer length of stay for patients undergoing rehabilitation after primary knee or hip arthroplasty.5 Pain is also associated with deterioration of both physical and mental quality of life scores as demonstrated for patients undergoing lower extremity orthopedic rehabilitation.6
In addition to affecting rehabilitation outcomes, pain in the inpatient rehabilitation setting is also of concern with regard to opioid use during rehabilitation. Evaluation of initial pain ratings among new admissions to inpatient rehabilitation facilities showed that increased pain intensity and total opioid consumption on the third day was associated with an increase in length of stay.5 These outcomes may also be impacted by pain disparities as a recent study demonstrated that non-Hispanic Black patients with traumatic brain injuries experienced greater pain and pain-related interference.7
While pain is associated with adverse outcomes for inpatient rehabilitation patients, there is a strong need for clarity regarding the prevalence of pain diagnoses in inpatient rehabilitation patients, clarification of pain diagnoses experienced by different patient groups, and identification of patients at-risk for pain in order to improve patient outcomes. To address these needs, this study examines the hypothesis that pain is frequently diagnosed in inpatient rehabilitation patients using a national database of rehabilitation patient discharges and literature-defined framework for identifying pain in large datasets. This national approach also allows for meaningful comparisons of pain diagnoses between impairment groups and identification of patient-level factors associated with the presence of a pain diagnosis. Examining comorbid pain diagnoses among inpatient rehabilitation patients sheds light on the extent of pain diagnoses, patient characteristics associated with pain, and will foster future interventions, including targeted approaches, to address pain symptoms and pain-related outcomes in these populations.
METHODS
Data Source and Study Design
This study is a retrospective review using deidentified inpatient rehabilitation discharges from the Uniform Data System for Medical Rehabilitation (UDSMR). The UDSMR data set consists of patient discharge information from over 80% of all inpatient rehabilitation facilities nationwide coded into the Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) and recorded in the software database by individual rehabilitation facilities typically for ensuring compliance with Centers for Medicare & Medicaid Services regulations and tracking outcomes.8 The IRF-PAI versions 1.3 and later contain information relating to a patient’s admission including impairment group assignment and International Classification of Diseases (ICD) codes for 3 etiologic diagnoses (the reason for admission), 25 comorbidity diagnoses, and 6 complication diagnoses.9,10 All data from this database are deidentified with data collection from inpatient rehabilitation discharges in accordance with the Helsinki Declaration of 1975, as revised in 1983. Given that the data are retrospective, deidentified, and lacking identifiable information from a national database, institutional review board approval was not required for this study. The UDSMR data for this study was obtained and used with permission from Netsmart Technologies, Inc. The service marks and trademarks associated with the FIM instrument are all owned by Netsmart Technologies, Inc.
Population and Inclusion/Exclusion Criteria
Acute inpatient rehabilitation patient discharges were identified in the USDMR database from January 2016 through December 2019. Patients were required to be at least 18 yrs old.
Study Variables and Methods
For the purposes of this study, each discharge in the UDSMR database was counted as one patient. Patient discharge diagnoses (including etiologic, comorbidity, and complication diagnoses) were screened for a pain diagnosis using a literature-established framework identifying International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) pain diagnoses based on the US National Pain Strategy.11 This framework was validated in three large healthcare systems across the ICD-9 to ICD-10 transition to be used for identifying common pain-related diagnoses.11 This framework broadly divides pain diagnoses into one of 13 pain clusters: back pain; neck pain; limb/extremity pain, joint pain, and nonsystemic, noninflammatory arthritic disorders; fibromyalgia; headache; orofacial, ear, and temporomandibular disorder pain; abdominal and bowel pain; urogenital, pelvic and menstrual pain; musculoskeletal chest pain; neuropathy; systemic disorders or diseases causing pain; other painful conditions; and fractures, contusions, sprains, and strains. These pain clusters were further subdivided into subgroups based on the same literature (shown in Supplemental Table 1, http://links.lww.com/PHM/C387).11
Demographic data collected from the UDSMR database included age at admission, gender, race, ethnicity, marital status, prehospital living situation (alone vs. with others), primary payer type, onset days, and admission Functional Independence Measure (FIM) scores (cognitive and motor).
Data from UDSMR patient discharges were stratified according to the presence or absence of at least one ICD-10-CM pain diagnosis using the ICD-10 codes in the literature framework for pain diagnoses. Patients were stratified by the 17 impairment groups corresponding to the primary inpatient rehabilitation diagnosis. Pain diagnoses were also reported by cluster and subgroup for each impairment group. For patient discharges with multiple ICD-10-CM pain diagnoses in a given cluster/subgroup, each patient discharge was only counted once per cluster and once per subgroup to ensure that multiple pain diagnoses per patient were not counted more than once in each cluster or subgroup.
Statistical Tests and Logistic Regression Model
Demographic and clinical data were compared between those with and without pain diagnoses using t tests (quantitative data) and χ2 tests (categorical data) using StataCorp 2021 (StataCorp LLC, Stata Statistical Software: version 17) with an a priori significance level of α = 0.05. Descriptive statistics were used to describe the prevalence of pain diagnosis by impairment groups/subgroups.
A multivariable logistic regression including all the covariates in Table 1, and listed also in Table 2, as well as impairment group was estimated using maximum likelihood analysis.12 The outcome or dependent variable was whether the patient had any pain codes. For the quantitative predictors (e.g., age or admission FIM), linearity was checked using lowess;13 as shown in Table 2, there were some nonlinear relationships, including quadratics for age and admission FIM and a piecewise linear form for onset days.
TABLE 1.
Comparison of demographic and clinical characteristics between patients with and without pain diagnoses in inpatient rehabilitation
| Pain Diagnosis Present? No Yes |
Total | P | ||
|---|---|---|---|---|
|
| ||||
| n (%) | 577,763 (30.0) | 1,347,239 (70.0) | 1,925,002 | |
| Age, mean years (SD) | 68.6 (14.9) | 69.3 (15.2) | 69.1 (15.1) | <0.001 |
| Female, n (%) | 261,278 (45.2) | 714,533 (53.0) | 975,811 (50.7) | <0.001 |
| Race/ethnicity, n (%) | <0.001 | |||
| White | 413,005 (74.3) | 1,046,039 (80.1) | 1,459,044 (78.4) | |
| Black | 84,058 (15.1) | 153,212 (11.7) | 237,270 (12.7) | |
| Hispanic/Latino | 35,445 (6.4) | 67,476 (5.2) | 102,921 (5.5) | |
| Other | 20,443 (3.7) | 34,396 (2.6) | 54,839 (2.9) | |
| Multiracial | 2747 (0.5) | 4972 (0.4) | 7719 (0.4) | |
| Married, n (%) | 69,558 (49.2) | 168,090 (47.0) | 237,648 (47.6) | <0.001 |
| Living alone, n (%) | 63,294 (24.7) | 184,687 (27.8) | 247,981 (26.9) | <0.001 |
| Primary payer, n (%) | <0.001 | |||
| Medicare | 401,480 (70.9) | 997,003 (75.4) | 1,398,483 (74.1) | |
| Medicaid | 42,827 (7.6) | 81,212 (6.1) | 124,039 (6.6) | |
| Commercial | 99,181 (17.5) | 184,692 (14.0) | 283,873 (15.0) | |
| Unreimbursed | 9907 (1.7) | 14,848 (1.1) | 24,755 (1.3) | |
| Worker’s Comp. | 2221 (0.4) | 13,528 (1.0) | 15,749 (0.8) | |
| Other | 10,532 (1.9) | 31,084 (2.4) | 41,616 (2.2) | |
| Onset days, mean (SD) | 11.9 (19.3) | 11.1 (21.3) | 11.3 (20.7) | <0.001 |
| Admission FIM cognitive, mean (SD) | 20.8 (7.4) | 22.8 (6.9) | 22.2 (7.1) | <0.001 |
| Admission FIM motor, mean (SD) | 37.2 (13.6) | 36.3 (12.3) | 36.6 (12.7) | <0.001 |
Adult inpatient rehabilitation patient discharges were identified in the UDSMR® database from January 2016 through December 2019. Patient discharge diagnoses (including etiologic, comorbidity, or complication diagnoses) were screened for pain diagnoses using a literature-established framework identifying ICD-10-CM pain diagnoses. Demographic and clinical data were compared between those with and without pain diagnoses using t tests (quantitative data) and χ2 tests (categorical data) using StataCorp 2021 (StataCorp LLC, Stata Statistical Software: version 17). The UDSMR data for this study was obtained and used with permission from Netsmart Technologies, Inc. The service marks and trademarks associated with the FIM instrument are all owned by Netsmart Technologies, Inc.
TABLE 2.
Logistic regression model examining characteristics associated with the presence of pain diagnosis
| Patient Characteristic | Odds Ratio | Standard Error | P | 95% Confidence Interval |
|---|---|---|---|---|
|
| ||||
| Age | 1.01 | 0.002 | <0.001 | 1.00–1.01 |
| Age, squared | 1.00 | 0.00001 | <0.001 | 1.00–1.00 |
| Onset to 8 d | 0.96 | 0.003 | <0.001 | 0.95–0.96 |
| Onset ≥ 8 d | 1.00 | 0.0002 | <0.001 | 1.00–1.00 |
| Sex (female) | 1.25 | 0.007 | <0.001 | 1.24–1.26 |
| Race/ethnicity | ||||
| Black | 0.82 | 0.02 | <0.001 | 0.77–0.87 |
| Hispanic/Latino | 0.80 | 0.04 | <0.001 | 0.72–0.88 |
| Other | 0.76 | 0.02 | <0.001 | 0.72–0.80 |
| Multiracial | 0.83 | 0.08 | 0.06 | 0.68–1.01 |
| Married | 1.04 | 0.008 | <0.001 | 1.02–1.05 |
| Living alone | 1.05 | 0.01 | <0.001 | 1.03–1.07 |
| Interaction of married and living alone | 0.93 | 0.02 | <0.001 | 0.90–0.96 |
| Primary payer | ||||
| Medicaid | 0.83 | 0.02 | <0.001 | 0.79–0.86 |
| Commercial | 0.75 | 0.007 | <0.001 | 0.74–0.77 |
| Unreimbursed | 0.70 | 0.03 | <0.001 | 0.65–0.76 |
| Worker’s comp. | 1.23 | 0.06 | <0.001 | 1.11–1.35 |
| Other | 0.96 | 0.02 | 0.11 | 0.91–1.01 |
| Admission FIM, total | 1.03 | 0.001 | <0.001 | 1.02–1.03 |
| Admission FIM, total, squared | 1.00 | 0.00001 | <0.001 | 1.00–1.00 |
| Impairment group | ||||
| Brain dysfunction | 1.32 | 0.02 | <0.001 | 1.29–1.36 |
| Neurologic conditions | 2.23 | 0.06 | <0.001 | 2.12–2.34 |
| Spinal cord dysfunction | 7.50 | 0.18 | <0.001 | 7.16–7.85 |
| Amputation | 4.11 | 0.10 | <0.001 | 3.91–4.31 |
| Arthritis | 12.97 | 0.87 | <0.001 | 11.38–14.78 |
| Pain syndromes | 11.13 | 1.03 | <0.001 | 9.29–13.34 |
| Orthopedic conditions | 6.56 | 0.14 | <0.001 | 6.29–6.84 |
| Cardiac disorders | 1.40 | 0.03 | <0.001 | 1.35–1.45 |
| Pulmonary disorders | 1.38 | 0.03 | <0.001 | 1.32–1.45 |
| Burns | 1.21 | 0.09 | 0.007 | 1.05–1.40 |
| Congenital deformities | 2.46 | 0.31 | <0.001 | 1.93–3.14 |
| Other disabling impairments | 2.46 | 0.10 | <0.001 | 2.27–2.67 |
| Major multiple trauma | 12.36 | 0.46 | <0.001 | 11.48–13.30 |
| Developmental disability | 1.78 | 0.81 | 0.20 | 0.73–4.36 |
| Debility | 1.89 | 0.03 | <0.001 | 1.83–1.96 |
| Medically complex conditions | 1.55 | 0.09 | <0.001 | 1.38–1.73 |
| Constant | 0.68 | 0.06 | <0.001 | 0.58–0.80 |
Logistic regression was completed examining characteristics associated with a pain diagnosis based on patient demographic and discharge data from inpatient rehabilitation facilities included in the UDSMR dataset. For onset days, linear splines were used separating this group into admissions up to 8 days and those ≥8 days. The baseline reference groups (constant) includes patients with the following demographic/clinical characteristics: male, White, not married, not living alone, Medicare as primary payer, and being in the stroke impairment group.
C statistic for the final model: 0.71.
Workers Comp, workers compensation.
Cluster-adjusted standard errors were used to account for the within facility clustering effect.14 No variable selection was performed. Two forms of overall assessment of the model were performed: first, its ability to discriminate between those with pain and those without pain was measured by the C statistic (area under the ROC curve); second, a calibration plot was used for internal calibration (comparison of predicted and observed probabilities).15,16
RESULTS
A total of 1,925,002 patients were identified in the UDSMR database with discharges from January 2016 through December 2019. Of these, 1,347,239 (70.0%) had at least one ICD-10 pain diagnosis. Demographic and clinical data are presented in Table 1.
Overall, female patients were more likely to have a diagnosis of pain (53.0% vs. 45.2%, P < 0.001). Patients without a pain diagnosis were more likely to be married (49.2% vs. 47.0%, P < 0.001). The distribution of race/ethnicities also differed between the patient groups with and without a pain diagnosis (P < 0.001) with more Black patients in the group without a pain diagnosis (15.1% vs. 11.7%) and more White patients in the group with a pain diagnosis (80.1% vs. 74.3%). Similarly, the distribution of primary payers also differed between the patient groups with and without a pain diagnosis (P < 0.001) with more patients with Medicare as a primary payer in the group with a pain diagnosis (75.4% vs. 70.9%) and more patients with commercial insurance in the group without a pain diagnosis (17.5% vs. 14.0%). Patients with a pain diagnosis had higher admission cognitive FIM score compared to patients without a pain diagnosis (mean = 22.8, SD = 6.9 vs. mean = 20.8, SD = 7.4; P < 0.001). Patients with a pain diagnosis had a lower admission FIM motor scores compared to patients without a pain diagnosis (mean = 36.3, SD = 12.3 vs. mean = 37.2, SD = 13.6; P < 0.001).
The total numbers of patients with a pain diagnosis per impairment group are shown in Table 3. Over half of the patients in all impairment groups had at least one pain diagnosis. A pain diagnosis was most frequent in the arthritis (94.1%), pain syndromes (93.0%), major multiple trauma (92.3%), orthopedic conditions (88.9%), and spinal cord dysfunction (88.5%) impairment groups. Pain diagnoses were least common, though still present in more than half of the patients, in the stroke (51.8%), burns (54.1%), and brain dysfunction (58.2%) impairment groups.
TABLE 3.
Frequency of patients with ICD-10-CM pain diagnosis by impairment group
| Impairment Group | Patients With a Pain Diagnosis, n (%) | Total Patients |
|---|---|---|
|
| ||
| Stroke | 236,336 (51.8) | 456,296 |
| Brain dysfunction | 128,098 (58.2) | 220,019 |
| Neurologic conditions | 185,060 (71.7) | 258,239 |
| Spinal cord dysfunction | 99,943 (88.5) | 112,974 |
| Amputation | 48,935 (80.3) | 60,975 |
| Arthritis | 7161 (94.1) | 7610 |
| Pain syndromes | 5399 (93.0) | 5803 |
| Orthopedic conditions | 370,687 (88.9) | 417,199 |
| Cardiac disorders | 54,294 (59.7) | 90,888 |
| Pulmonary disorders | 17,600 (60.8) | 28,944 |
| Burns | 1393 (54.1) | 2573 |
| Congenital deformities | 403 (73.3) | 550 |
| Other disabling impairments | 13,372 (72.6) | 18,424 |
| Major multiple trauma | 51,025 (92.3) | 55,299 |
| Developmental disability | 14 (66.7) | 21 |
| Debility | 120,061 (67.9) | 176,849 |
| Medically complex conditions | 7458 (60.4) | 12,339 |
Adult inpatient rehabilitation patient discharges were identified in the UDSMR database from January 2016 through December 2019. Patient discharge diagnoses (including etiologic, comorbidity, or complication diagnoses) were screened for pain diagnoses using a literature-established framework identifying ICD-10-CM pain diagnoses. Patient discharges were then stratified by the presence or absence of at least one pain diagnosis for each impairment group. UDSMR impairment groups are sorted based on the coding from the CMS IRF-PAI versions 1.3 and later. The UDSMR data for this study was obtained and used with permission from Netsmart Technologies, Inc.
The percentage of patients with a diagnosis in each pain cluster per impairment group is shown in Table 4. Of note, patients with multiple pain diagnoses are counted once in each applicable cluster. Overall, the cluster corresponding with limb/extremity and joint pain was prevalent in all impairment groups (range 9.0%–43.5%) with highest prevalence in the arthritis impairment group. Neuropathy was the second most prevalent pain cluster (range 3.7%–30.3%) with highest prevalence in the amputation impairment group. The next most frequently diagnosed pain clusters included other painful conditions (range 4.8%–17.8%); fractures, contusions, sprains, and strains (range 1.2%–63.2%); and back pain (range 2.1%–30.8%). The least commonly diagnosed pain clusters across impairment groups included orofacial, ear, and temporomandibular disorder pain (range 0–0.2%); urogenital, pelvic, and menstrual pain (range 0.2–0.5%); and fibromyalgia (range 0.4–1.3%). The percentage of patients with a diagnosis in each pain subgroup per impairment group is shown in Supplemental Table 1, http://links.lww.com/PHM/C387.
TABLE 4.
Percentage of patients in each pain cluster by impairment group
| Stroke | TBI | Neurologic | SCI | Amputation | Arthritis | Pain | Orthopedic | Cardiac | Pulm | Burn | Cong | Other | Multi traum | Devel | Debility | Complex | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||||||||||
| Abdominal and bowel pain | 2.3 | 2.4 | 3.1 | 1.6 | 1.1 | 1.7 | 2.2 | 1.9 | 2.8 | 3.6 | 1.5 | 3.6 | 3.3 | 1.3 | 4.9 | 3.8 | |
| Back pain | 4.6 | 5.0 | 10.6 | 16.4 | 2.1 | 7.1 | 30.8 | 7.1 | 5.1 | 6.6 | 2.4 | 8.9 | 5.4 | 2.8 | 9.5 | 6.5 | 5.4 |
| Fibromyalgia | 0.6 | 0.7 | 0.9 | 0.8 | 0.5 | 1.3 | 1.3 | 0.8 | 0.6 | 0.9 | 0.5 | 0.5 | 0.8 | 0.4 | 0.9 | 0.7 | |
| Fractures, contusions, sprains and strains | 2.3 | 9.7 | 3.2 | 6.5 | 1.2 | 4.0 | 6.2 | 31.2 | 2.2 | 3.0 | 4.7 | 1.8 | 3.1 | 63.2 | 14.3 | 3.8 | 3.8 |
| Headache | 1.4 | 2.0 | 1.1 | 1.0 | 0.3 | 0.6 | 0.8 | 0.8 | 0.5 | 0.7 | 0.7 | 3.3 | 1.6 | 1.0 | 0.8 | 0.7 | |
| Limb/extremity pain, joint pain, and nonsystemic, noninflammatory arthritic disorders | 20.3 | 17.2 | 22.1 | 18.3 | 26.4 | 43.5 | 23.9 | 25.0 | 21.8 | 19.5 | 16.4 | 24.0 | 19.0 | 9.0 | 23.8 | 22.4 | 17.6 |
| Musculoskeletal chest pain | 1.0 | 0.9 | 1.0 | 0.6 | 0.4 | 0.5 | 0.7 | 0.6 | 2.3 | 1.5 | 0.4 | 1.1 | 1.0 | 0.7 | 1.1 | 1.0 | |
| Neck pain | 1.6 | 2.0 | 1.8 | 12.9 | 0.3 | 1.3 | 3.2 | 1.1 | 0.9 | 1.0 | 0.7 | 2.0 | 1.3 | 1.7 | 1.2 | 1.2 | |
| Neuropathy | 9.3 | 9.1 | 15.7 | 11.1 | 30.3 | 9.5 | 8.5 | 7.7 | 13.8 | 11.5 | 11.0 | 8.9 | 20.6 | 3.7 | 14.3 | 13.8 | 11.5 |
| Orofacial, ear, and temporomandibular disorder pain | 0.1 | 0.2 | 0.1 | 0.1 | 0.0 | 0.1 | 0.0 | 0.0 | 0.1 | 0.1 | 0.1 | 0.1 | 0.1 | 0.1 | 0.1 | ||
| Other painful conditions | 6.1 | 7.2 | 9.4 | 17.5 | 15.8 | 13.2 | 13.2 | 10.4 | 7.4 | 9.2 | 14.3 | 17.8 | 13.4 | 7.7 | 4.8 | 9.5 | 11.8 |
| Systemic disorders or diseases causing pain | 1.7 | 1.7 | 2.3 | 1.5 | 1.6 | 11.3 | 2.0 | 2.1 | 2.1 | 3.0 | 0.9 | 1.1 | 2.5 | 0.6 | 2.5 | 2.3 | |
| Urogenital, pelvic and menstrual pain | 0.2 | 0.2 | 0.3 | 0.2 | 0.3 | 0.2 | 0.2 | 0.2 | 0.3 | 0.2 | 0.5 | 0.2 | 0.4 | 0.3 | 0.4 | 0.4 | |
Adult inpatient rehabilitation patient discharges were identified in the UDSMR database from January 2016 through December 2019. Patient discharge diagnoses (including etiologic, comorbidity, or complication diagnoses) were screened for pain diagnoses using a literature-established framework identifying ICD-10-CM pain diagnoses which stratifies pain diagnoses into diagnostic clusters. Patient discharges were then stratified by the presence or absence of a pain diagnosis for each impairment group. Of note, patients could have more than one pain diagnosis placing them in more than one cluster. Each patient is only counted once per cluster. UDSMR impairment groups are sorted based on the coding from the CMS IRF-PAI versions 1.3 and later. The UDSMR data for this study was obtained and used with permission from Netsmart Technologies, Inc.
Cardiac, cardiac disorders; Complex, medically complex; Cong, congenital deformities; Devel, developmental disability; Multitraum, major multiple trauma; Neurologic, neurologic conditions; Orthopedic, orthopedic conditions; Other, other disabling impairments; Pain, pain syndromes; Pulm, pulmonary disorders; SCI, spinal cord dysfunction; TBI, brain dysfunction.
In the logistic regression model, multiple patient-related factors were significantly associated with a greater and lesser odds of a pain diagnosis (Table 2) with calibration plot shown in Supplemental Figure 1, http://links.lww.com/PHM/C388 (C statistic for the final model: 0.71). The odds of a pain diagnosis were higher for patients of female sex, those with worker’s compensation as a primary payer, and those with diagnoses in the following impairment groups: arthritis, major multiple trauma, and pain syndromes. Similarly, odds of a pain diagnosis were lower for patients who were of Black, Hispanic/Latino, and other race/ethnicity and those with Medicaid, commercial, and unreimbursed primary payers. The baseline group had the lowest odds of a pain diagnosis, representing patients who were male, White, unmarried, not living alone, using Medicare as a primary payer, and in the stroke impairment group.
DISCUSSION
While pain is regularly managed in inpatient rehabilitation patients, the prevalence of pain, pain diagnoses, and patient characteristics associated with pain in this setting has not been examined on the national level. This study reports the frequency, types of pain, and characteristics associated with pain experienced by patients across all impairment groups using a large national dataset with over 1.9 million inpatient rehabilitation patient discharges. Overall, over half of the patients in each impairment group had at least one ICD-10 diagnosis of a painful condition. A pain diagnosis was most common in the arthritis, pain syndromes, and major multiple trauma impairment groups and least common in the stroke, burn, and brain dysfunction impairment groups. Using a logistic regression model, the odds of a pain diagnosis was higher for patients who were female, having worker’s compensation as a primary payer, and those in the arthritis, major multiple trauma, and pain syndromes impairment groups. The most common pain diagnoses included codes corresponding to limb/extremity/joint pain, fractures, and neuropathy with variation between impairment groups for each pain cluster.
Using a national database, the results of this study suggest that pain diagnoses are common in inpatient rehabilitation and helps clarify the existing literature by examining pain diagnoses between impairment groups and demonstrating patient-level characteristics associated with greater (or lesser) odds of a pain diagnosis. Existing studies examining pain inpatient rehabilitation patients have smaller patient populations with heterogeneity between studies limiting broader analysis. Data from the SCIRehab Study (n = 1357) suggest that 97% of patients undergoing rehabilitation for a traumatic spinal cord injury experience pain at least once during rehabilitation, with notable variations in average pain intensity between patients.2 Conversely, of 584 patients undergoing rehabilitation for neurologic diagnoses in Germany, 149 had clinically relevant pain at the beginning of their inpatient stay, which increased to 164 at the end of rehabilitation.3 A total of 159 patients required pain medication in the German study.3 In a different study examining 46 patients undergoing inpatient rehabilitation after a stroke, 37% reported shoulder pain.17 While not examined here, the existing literature also indicates differences between patient populations in terms of the severity of pain. For example, a single-center study in Chicago, IL, found that 388 of 1025 admitted patients had at least one order for an opioid medication, potentially suggesting a high prevalence of more intense pain.18 Overall, these studies indicate that pain is frequently diagnosed in inpatient rehabilitation patients but are limited to smaller patient populations and fewer impairment groups with some heterogeneity between studies likely due to inclusion criteria. This may also be due to the different types of pain (such as nociceptive, neuropathic, and visceral) affecting patients which can be exacerbated by comorbid psychiatric conditions. This present work also suggests that the prevalence of pain in inpatient rehabilitation patients may be similar to the acute care hospital setting where pain is estimated to affect 37.7% to 84% of patients.1
While pain was prevalent across all impairment groups, there were differences in pain diagnoses between impairment groups. For example, patients in the orthopedic and multiple trauma groups had the highest prevalence of pain diagnoses in the fractures, contusions, sprains, and strains cluster while these diagnoses were the least common in the amputation impairment group. Patients in the pain syndromes and spinal cord dysfunction impairment groups had the highest prevalence of pain diagnoses in the back pain cluster compared to the amputation and major multiple trauma impairment groups. These comparisons suggest correspondence between impairment group and pain diagnosis.
This study also revealed that patients coded as Black, Hispanic/Latino, or other races had a lower odds of a pain diagnosis. Racial disparities resulting in differences in pain severity, impact, and unequal access to pain treatment are well-documented in the literature.19–21 In terms of the impact of chronic pain, Black patients are more likely to report greater pain severity and, as a result, greater pain-related disability than White patients.19,22 Despite this, across treatment settings, Black and Hispanic patient populations are less likely to receive opioid prescriptions for pain management than White patients.19,23 The reasons for a lower odds of a pain diagnosis for these groups of patients is not clear from this present study but highlights the ongoing need for improved pain assessment and management to avoid unequal care.
There are several limitations to the present work. While the USDMR database provides data for individual patient discharges during the specified timeframe, these results are deidentified. Thus, it is impossible to identify patients with multiple admissions who may be included more than once in this dataset. The diagnoses of pain are inferred indirectly using ICD-10 diagnoses corresponding to painful conditions and limited by the 36 ICD-10 codes that can be reported for each patient in the IRF-PAI. These codes may also be entered for purposes of reimbursement by facilities. The severity of pain experienced by each patient and further details about the pain diagnoses themselves including duration, quality, severity, treatment, and comorbid behavioral health stressor(s) cannot be assessed from the available data. Thus, the chronicity of pain including the diagnosis of pain as acute versus chronic was unable to be ascertained from this data. As the database lacks information on pain treatment including analgesic use, the impact of individual pain diagnoses on patient management could not be ascertained. While this study uses a literature-established framework for identifying ICD-10 codes for pain in large datasets, some pain diagnoses may be excluded. For example, manual review of the etiologic diagnoses available for patients without pain diagnoses in the pain syndromes impairment group showed a very small minority with ICD-10 diagnoses potentially causing back pain that were excluded from the literature framework (data not shown). Patients may also fail to express pain resulting in a missed diagnosis. However, a post-hoc analysis of the data did not demonstrate notable differences in pain prevalence between patients with and without a language disorder in the stroke and brain injury impairment groups (data not shown). The literature framework utilized here also could not differentiate between expected posttraumatic or postsurgical pain versus chronic pain reflecting development of sensitization. Overall, the prevalence of pain in inpatient rehabilitation patients is likely even higher than demonstrated here.
It is also important to note that a separate pain diagnosis may not be made when pain is inferred to be related to the admitting diagnosis or when separate therapeutic interventions are not prescribed resulting in lack of ICD-10-CM coding by facilities. For example, patients in the amputee impairment group had the highest prevalence of diagnoses of neuropathy (30.3%) compared to patients in the spinal cord dysfunction impairment group (11.1%), which differs from prevalence estimates in the literature. For neuropathic pain, the literature for patients with spinal cord injuries estimates that 56% of patients have neuropathic pain.24 The prevalence of neuropathic pain from amputation is less clear, but one meta-analysis examining phantom limb pain suggests that it affects 64% of patients.25 However, patients with amputations were coded as having pain in the cluster relating to limb/extremity pain. Similarly, the prevalence of headache in the traumatic brain dysfunction impairment group is lower in this study compared to outpatient visits suggesting that sources of mild pain or pain related to admitting diagnosis may not always be given a separate pain diagnosis.26 Lastly, the burns impairment group had a lower prevalence of pain compared to other impairment groups, though a higher odds compared to the baseline group in the regression model. While the literature framework appears to have excluded burn diagnoses, pain related to burns (such as from dressing changes or limb repositioning) may also be uncoded. In the literature, approximately 52% of burn survivors report chronic pain with 66% of those with pain reporting that it interferes with rehabilitation after injury.27
While this study was able to demonstrate that patients with a pain diagnosis had slightly higher cognitive FIM scores and lower motor FIM scores than those without pain at admission, the impact of pain on rehabilitation functional outcomes remains to be examined. A diagnosis of pain may affect functional outcomes. In the SCIRehab study of patients with spinal cord injuries, those with severe pain (28.4%) spent less time in physical therapy and therapeutic recreation, and with social work.2 These patients also missed more hours per week of occupational therapy sessions compared to those with less severe or no pain and reported the most overall therapy sessions impacted by pain.2 The effect of pain on rehabilitation outcomes must be examined more broadly across all impairment groups. Determination of pain diagnoses impacting functional outcomes could serve as the foundation for development of a comprehensive strategy to address pain management in rehabilitation patients, such as that proposed for patients with cancer pain.28 While interventional procedures are often done in the outpatient setting due to the need for previous authorizations and reimbursement limitations in the inpatient setting, coordination of care to include outpatient interventional pain management could be of great benefit to patients undergoing inpatient rehabilitation.28
Using data from over a million inpatient rehabilitation discharges, this study establishes pain as a common diagnosis affecting 70% of patients in the rehabilitation setting. A pain diagnosis was common among all impairment groups, affecting over half of patients in each group with the highest prevalence in the arthritis and pain syndromes impairment groups. This high prevalence of painful conditions, particularly during the opioid epidemic, calls for a renewed focus on pain management for patients undergoing rehabilitation. In addition to a renewed focus on the individual level, development of a strategic approach at the institutional level and consideration of pain specialty referral is also needed.
Supplementary Material
What Is Known
While it is known that inpatient rehabilitation patients experience pain, there are limited data summarizing the prevalence of pain and characteristics associated with pain diagnoses in inpatient rehabilitation patients.
What Is New
This study demonstrates the prevalence of pain diagnoses and characteristics associated with pain diagnoses in inpatient rehabilitation patients using data from over 1.9 million inpatient rehabilitation patient discharges in a national database. This study also allows for the direct comparison of pain diagnoses between impairment groups.
Acknowledgments
The contents of this manuscript were developed under grants from the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR grant numbers 90DPBU0008, 90DPTB0011-05-00, and 90SIMS0017). NIDILRR is a Center within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The contents of this manuscript represent the view of the authors and do not necessarily reflect the position or policy of the NIDILRR, ACL, HHS, the U.S. Department of Veterans Affairs, the National Institutes of Health, nor the United States Government. Dr. Zafonte received royalties from Springer/Demos publishing for serving as co-editor of the text Brain Injury Medicine. He is a member of the editorial board of the Journal of Neurotrauma and Frontiers in Neurology. Dr. Zafonte serves on the Scientific Advisory Board of Myomo, Nanodiagnostics, and Onecare.ai. He also evaluates patients in the MGH Brain and Body-TRUST Program which is funded by the National Football League Players Association. The authors otherwise do not have any other competing interests, funding, or financial benefits to disclose.
Footnotes
Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
Author Disclosures: An earlier version of this research was previously published as an abstract for poster presentation at Physiatry ‘23, annual meeting of the Association of Academic Physiatrists, in Anaheim, CA. The presenting author was ZAC.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.ajpmr.com).
Contributor Information
Zachary A. Curry, Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Rehabilitation Outcomes Center at Spaulding, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Department of Physical Medicine and Rehabilitation, Massachusetts General Hospital, Boston, Massachusetts; Harvard Medical School, Boston, Massachusetts.
Michael N. Andrew, Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Rehabilitation Outcomes Center at Spaulding, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Department of Physical Medicine and Rehabilitation, Massachusetts General Hospital, Boston, Massachusetts; Harvard Medical School, Boston, Massachusetts.
Michael C. Chiang, Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Rehabilitation Outcomes Center at Spaulding, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Department of Physical Medicine and Rehabilitation, Massachusetts General Hospital, Boston, Massachusetts; Harvard Medical School, Boston, Massachusetts.
Richard Goldstein, Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Rehabilitation Outcomes Center at Spaulding, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts.
Ross Zafonte, Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Rehabilitation Outcomes Center at Spaulding, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Department of Physical Medicine and Rehabilitation, Massachusetts General Hospital, Boston, Massachusetts; Harvard Medical School, Boston, Massachusetts.
Colleen M. Ryan, Harvard Medical School, Boston, Massachusetts; Department of Surgery, Massachusetts General Hospital, Boston, Massachusetts; Department of Surgery, Shriners Children’s, Boston, Massachusetts.
Brian C. Coleman, Pain Research, Informatics, Multimorbidities, and Education (PRIME) Center, VA Connecticut Healthcare System, West Haven, Connecticut; Department of Emergency Medicine, Yale School of Medicine, New Haven, Connecticut.
Jeffrey C. Schneider, Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Rehabilitation Outcomes Center at Spaulding, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts; Department of Physical Medicine and Rehabilitation, Massachusetts General Hospital, Boston, Massachusetts; Harvard Medical School, Boston, Massachusetts.
Data Availability Statement:
The data that support the findings of this study are available from the corresponding author but restrictions apply to the availability of these data, which were used under data use agreement for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Netsmart Technologies, Inc.
REFERENCES
- 1.Gregory J, McGowan L: An examination of the prevalence of acute pain for hospitalised adult patients: a systematic review. J Clin Nurs 2016;25(5–6):583–98 [DOI] [PubMed] [Google Scholar]
- 2.Zanca JM, Dijkers MP, Hammond FM, et al. : Pain and its impact on inpatient rehabilitation for acute traumatic spinal cord injury: analysis of observational data collected in the SCIRehab study. Arch Phys Med Rehabil 2013;94(4 Suppl):S137–44 [DOI] [PubMed] [Google Scholar]
- 3.Timm A, Knecht S, Florian M, et al. : Frequency and nature of pain in patients undergoing neurorehabilitation. Clin Rehabil 2021;35:145–53 [DOI] [PubMed] [Google Scholar]
- 4.Amtmann D, Bamer AM, McMullen K, et al. : Pain across traumatic injury groups: a National Institute on Disability, Independent Living, and Rehabilitation Research model systems study. J Trauma Acute Care Surg 2020;89:829–33 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sahota B, Alavinia SM, Kumbhare D, et al. : A retrospective study of the association between pain intensity and opioid use with length of stay during musculoskeletal inpatient rehabilitation after primary knee and hip arthroplasty. PM R 2020;12:462–9 [DOI] [PubMed] [Google Scholar]
- 6.Padua L, Aprile I, Cecchi F, et al. : Pain in postsurgical orthopedic rehabilitation: a multicenter study. Pain Med 2012;13:769–76 [DOI] [PubMed] [Google Scholar]
- 7.Sander AM, Williams M, Loyo K, et al. : Disparities in chronic pain experience and treatment history among persons with traumatic brain injury: a traumatic brain injury model systems study. J Head Trauma Rehabil 2023;38:125–36 [DOI] [PubMed] [Google Scholar]
- 8.Rehab Inpatient. Uniform Data System for Medical Rehabilitation‚ a division of UB Foundation Activities‚ Inc Available at: https://www.udsmr.org/products/inpatient-rehab. Accessed June 2023 and December 2023
- 9.Department of Health and Human Services Center for Medicare & Medicaid Services. Patient assessment instrument for use in an inpatient rehabilitation facility. OMB No. 0938–0842. Available at: https://www.cms.gov/medicare/medicare-fee-for-service-payment/inpatientrehabfacpps/downloads/irf-pai_for_fy2015_final.pdf. Accessed June 2023
- 10.Department of Health and Human Services Centers for Medicare & Medicaid Services. IRF-PAI versions 1.3 and later. Published online May 2, 2022. Available at: https://www.cms.gov/medicare/medicare-fee-for-service-payment/inpatientrehabfacpps/irfpai. Accessed June 2023
- 11.Mayhew M, DeBar LL, Deyo RA, et al. : Development and assessment of a crosswalk between ICD-9-CM and ICD-10-CM to identify patients with common pain conditions. J Pain 2019;20:1429–45 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hosmer DW, Lemeshow S, Sturdivant RX: Applied Logistic Regression, 3rd ed. John Wiley & Sons, Inc., Hoboken, New Jersey. 2013 [Google Scholar]
- 13.Cleveland WS: Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979;74:829–36 [Google Scholar]
- 14.Cameron AC, Miller DL: A practitioner’s guide to cluster-robust inference. J Hum Resour 2014;50:317–72 [Google Scholar]
- 15.Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29–36 [DOI] [PubMed] [Google Scholar]
- 16.Austin PC, Steyerberg EW: Graphical assessment of internal and external calibration of logistic regression models by using loess smoothers. Stat Med 2014;33:517–35 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Dromerick AW, Edwards DF, Kumar A: Hemiplegic shoulder pain syndrome: frequency and characteristics during inpatient stroke rehabilitation. Arch Phys Med Rehabil 2008;89: 1589–93 [DOI] [PubMed] [Google Scholar]
- 18.Glinka Przybysz A, Khudeira Z, Khudeira S, et al. : Opioid prescribing and utilization during acute inpatient rehabilitation admissions. Pain Med 2021;22:3089–91 [DOI] [PubMed] [Google Scholar]
- 19.Morales ME, Yong RJ: Racial and ethnic disparities in the treatment of chronic pain. Pain Med 2021;22:75–90. doi: 10.1093/pm/pnaa427 [DOI] [PubMed] [Google Scholar]
- 20.Wyatt R: Pain and ethnicity. Virtual Mentor 2013;15:449–54 [DOI] [PubMed] [Google Scholar]
- 21.Edwards RR, Tan CO, Dairi I, et al. : Race differences in pain and pain-related risk factors among former professional American-style football players. Pain 2023;164:2370–9. Published online June 9, 2023:Online ahead of print [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Green CR, Hart-Johnson T: The impact of chronic pain on the health of black and white men. J Natl Med Assoc 2010;102:321–31 [DOI] [PubMed] [Google Scholar]
- 23.Meghani SH, Byun E, Gallagher RM: Time to take stock: a meta-analysis and systematic review of analgesic treatment disparities for pain in the United States. Pain Med 2012;13:150–74 [DOI] [PubMed] [Google Scholar]
- 24.Felix ER, Cardenas DD, Bryce TN, et al. : Prevalence and impact of neuropathic and nonneuropathic pain in chronic spinal cord injury. Arch Phys Med Rehabil 2022;103:729–37 [DOI] [PubMed] [Google Scholar]
- 25.Limakatso K, Bedwell GJ, Madden VJ, et al. : The prevalence and risk factors for phantom limb pain in people with amputations: a systematic review and meta-analysis. PLoS One 2020;15:e0240431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ashina H, Dodick DW, Barber J, et al. : Prevalence of and risk factors for post-traumatic headache in civilian patients after mild traumatic brain injury: a TRACK-TBI Study. Mayo Clin Proc 2023;98:1515–26. Published online July 21, 2023:S0025–6196(23)00113–1 [DOI] [PubMed] [Google Scholar]
- 27.Dauber A, Osgood PF, Breslau AJ, et al. : Chronic persistent pain after severe burns: a survey of 358 burn survivors. Pain Med 2002;3:6–17 [DOI] [PubMed] [Google Scholar]
- 28.Pugh TM, Squarize F, Kiser AL: A comprehensive strategy to pain management for cancer patients in an inpatient rehabilitation facility. Front Pain Res (Lausanne) 2021;2:688511. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author but restrictions apply to the availability of these data, which were used under data use agreement for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Netsmart Technologies, Inc.