Pre-Diagnosis Pain in Patients With Pancreatic Cancer Signals the Need for Aggressive Symptom Management

Abstract Objective This study assessed the impact of pancreatic cancer (PC) pain on associated symptoms, activities, and resource utilization from 2016 to 2020 in an online patient registry. Patients and Methods Responses from PC patient volunteers (N = 1978) were analyzed from online surveys in a cross-sectional study. Comparisons were performed between PC patient groups reporting, (1) the presence vs. absence of pre-diagnosis PC pain, (2) high (4-8) vs. low (0-3) pain intensity scores on an 11-point numerical rating scale (NRS), and (3) year of PC diagnosis (2010-2020). Descriptive statistics and all bivariate analyses were performed using Chi-square or Fisher’s Exact tests. Results PC pain was the most frequently reported pre-diagnosis symptom (62%). Pre-diagnostic PC pain was reported more frequently by women, those with a younger age at diagnosis, and those with PC that spread to the liver and peritoneum. Those with pre-diagnostic PC pain vs. those without reported higher pain intensities (2.64 ± 2.54 vs.1.56 ± 2.01 NRS mean ± SD, respectively, P = .0039); increased frequencies of post-diagnosis symptoms of cramping after meals, feelings of indigestion, and weight loss (P = .02-.0001); and increased resource utilization in PC pain management: (ER visits N = 86 vs. N = 6, P = .018 and analgesic prescriptions, P < .03). The frequency of high pain intensity scores was not decreased over a recent 11-year span. Conclusions PC pain continues to be a prominent PC symptom. Patients reporting pre-diagnosis PC pain experience increased GI metastasis, symptoms burden, and are often undertreated. Its mitigation may require novel treatments, more resources dedicated to ongoing pain management and surveillance to improve outcomes.


Introduction
Pancreatic cancer (PC) continues to have a dismal prognosis and high symptom burden from the disease and from side effects of treatment to reduce the tumor. 1,27][8] Recently published studies have reported PC pain's association with patient age, cancer stage, survival, Health-Related Quality of Life (HRQoL), and physical function. 7,91][12] Higher pain levels are associated with decreased caloric intake, potentially jeopardizing treatment eligibility and immunocompetence, 7,9,13 further decreasing patient HRQoL and overall survival. 6,13,14he Pancreatic Cancer Action Network (PanCAN) developed several programs to improve patient and caregiver access to care and support under its Patient Services umbrella.In 2016, PanCAN introduced a PC patient registry and survey.Patients were invited to join the patient registry and complete its online PC Survey, allowing the scientific community insight through assessment of symptoms and patient-reported outcomes (PRO) impacted by PC diagnosis and the patient's

Variables Description
Patient Information Survey Patient demographic items were embedded in the patient information survey and included questions or variables such as age at survey completion, age at PC diagnosis, gender, race, and ethnicity.

PC Experience Survey
The PC experience survey included variables such as pre-diagnosis and-post-diagnosis symptoms, characteristics of the PC diagnosed (type, stage, and affected organs), and types of treatment received.

Health Assessment Survey
The Health Assessment survey included 5 health domain variables: physical functioning, pain interference, anxiety, depression, and fatigue, comprised of 4 statement items per domain for rating abilities to participate in activities and social roles.Variables were derived from PROMIS (Patient-Reported Outcomes Measurement Information System) using portions of the PROMIS-29 v2.0 survey. 23,24Variables reflecting functioning and mood were classified from multiple categories into 2 distinct categories of either "Large" impairment or "Little" impairment by combining possible choices, similar to the categories previously reported 10 Pain Intensity/Rating Item The pain intensity item embedded in the Health Assessment Survey used an 11-point NRS for pain scoring, based on a scale from 0 = "No pain," to 10 = "Worst imaginable pain." 3,25he question asked of respondents was, "How would you rate your pain on average?"

Pain Management Survey
The drop-down pain management survey included variables such as resource utilization (emergency room (ER) visits and hospitalizations for pain management), doctors/other providers (referred here as HCPs), and pain medications and therapies recommended/prescribed for PC pain management.Also queried were items regarding PC pain characteristics of physical locations, pain types, and descriptors.

Study Population
Survey participants were eligible for this study if: (1) they were diagnosed with PC and completed the Registry for themselves and (2) they responded to a statement item (yes or no) from the PC Experience survey which queried whether participants had pre-diagnosis symptoms of "abdominal and/ or back (A/B) pain."The inclusion/exclusion criteria used for this study between 2016 and 2020 are shown in Fig. 1.Approximately 40 000 PC patients or their representatives accessed PanCAN Patient Services and N = 2247 voluntarily joined the online PRO Registry, responding to most or part of the survey items.PC patients who had no or minimal (<70%) responses to the relevant surveys were excluded and those who answered for themselves, N = 1992; were included for The Oncologist, 2023, Vol. 28, No. 12 e1187 consideration of study participation.Respondents were further selected those who answered whether they experienced abdominal and/or back pain, indicating PC pain, prior to their diagnosis of pancreatic cancer, N = 1978.Hence, the data presented represent single, (nonredundant), patient (not surrogate) responses from PC patient volunteers.

Statistical Analysis
All the analyses were conducted in SAS 9.4 English version and a P-value ≤ .05 was considered significant.Bivariate analysis using Chi-square or Fisher's exact test for categorical variables and Student's t test for continuous variables were performed to assess the association of reported pre-diagnosis abdominal and/or back pain symptoms (presence vs. absence) and variables in the surveys related to patient PC experience, health assessment, and pain management.
A sensitivity analysis using the NRS pain variable assessed associations between pain severity and variables in the selected surveys.NRS pain scores were categorized into 2 groups."Low" NRS pain scores 0-3, no to mild pain, were deemed to have usually manageable pain intensities with little impact on daily activities or mood and "High" NRS pain scores 4-8, moderate to severe pain, were deemed very likely to impact daily activities or mood for most people. 3,10,25dditional sensitivity analyses assigned patients to 2 groups, evaluating the associations between the year of PC diagnosis; years prior to 2018 vs. years 2018-2020.Variables compared were days from diagnosis to survey completion, the presence vs. absence of pre-diagnosis PC pain, NRS pain category variables, and PC treatment received to assess response frequency changes in pain intensities between years of survey completion and treatment trends.In addition, a subgroup analysis to compare patients who responded to the survey within 3 months of diagnosis vs. those who responded greater than 19 months after diagnosis was performed.

Patient Information
The total number of responses to items from patient information and PC experience surveys, including patient demographics and characteristics of the patients' PC, is shown in Table 1.The average age at PC diagnosis was 60.5 ± 11.0 years and the survey was completed on average 2 years later.There were approximately equal numbers of male and female respondents and race was predominately White with non-Hispanic ethnicity.
PC patients were asked to identify the PC-related pre-diagnosis and post-diagnosis symptoms they experienced from lists of possible choices.The top 3 symptoms reported in the pre-diagnosis period were abdominal and/or back pain (61.8%), weight loss or loss of appetite (47.4%), and digestive problems (46.6%).

Effect of Pre-Diagnosis Pain on PC Experience
Respondents were placed into 2 groups based on their responses to "Before my diagnosis, I had ….Abdominal and/ or back pain": those with pre-diagnosis pain and those with no pre-diagnosis pain (Table 1).Pre-diagnosis PC associations were not significant for tumor type reported, with similar frequencies for adenocarcinoma (78.5%) and neuroendocrine (6.5%).PC pain was reported in all PC tumor stages at diagnosis (resectable:3.2%;borderline resectable:78.8%;locally advanced: 11.3% and metastatic: 5.9%).Significantly increased frequencies of PC tumor spread to liver (38.5%), peritoneum (67.2%), and other organs (10.2%) associated with the presence of pre-diagnosis pain.
The frequencies of associated pre-diagnosis symptoms were significantly increased in the presence vs. absence of pre-diagnosis PC pain for changes in bowel movements, digestive problems, weight loss or loss of appetite, fatigue, and depression, and significantly decreased for Jaundice and the associated symptoms of dark urine and itching.Post-diagnosis symptom comparisons found significantly increased frequencies in the presence of pre-diagnosis PC pain with weight loss, cramping after meals, and feelings of indigestion, (eg, symptoms that manifest with pain or discomfort in the upper GI region) and significantly decreased frequencies of the item, have not experienced any of these symptoms.Response comparisons were not significant between the presence vs. absence of pre-diagnosis pain and remaining post-diagnosis symptoms related to pronounced GI disruption.The presence of pre-diagnosis PC pain was associated with longer duration of symptoms before PC diagnosis, documented by significantly decreased frequencies for <30 days, while reports of durations between 1 month to over 1 year were consistently higher.
Respondents were also queried on their treatment for PC, of which 63.0% of the total respondents were currently receiving treatment.Treatments over the course of the disease included chemotherapy (80.2%), surgery (46.2%), radiation (25.7%), and enrolled in a clinical trial (12.6%).Patients reporting the presence of pre-diagnosis PC pain had significantly decreased frequencies of radiation and surgery treatments, with no differences for those indicating they received chemotherapy or enrolled in a clinical trial.PC patients who had no responses to the patient information or answered less than 70% of the items in the PC experience surveys were excluded from the study (N = 146, 6.5%, top, level 1, right side).Patients who responded "Yes" or "No" to the survey item "Before my diagnosis I had: abdominal and/or back pain, " (pre-diagnosis abdominal/back pain, N = 1978, mid-level 3, left side) were included.N = number of participants in each population, n = number of participants in subcategory.

e1188
The Oncologist, 2023, Vol. 28, No. 12 Table 1.Characteristics of respondents reporting the presence vs. absence of pre-diagnosis pain.The responses of the PC patient volunteers to the PanCAN patient information and PC experience survey items in the online PanCAN survey (column 1), collected over the 5-year time frame between 2016 and 2020, are shown.Comparisons were generated between PC patients who reported "Yes" or presence of pre-diagnosis abdominal and/or back pain (A/B pain, or pre-diagnosis PC pain) (Column 2) and those who reported "No" or absence of pre-diagnosis abdominal and/or back pain (column 3).The total number of respondents/item are shown (column 4).The P-values are shown between comparisons of frequencies (%) of the "Yes" vs. "No" responses (column 5).Patients reporting the presence vs. absence of pre-diagnosis PC pain were compared to patient responses to statements of functioning and mood (N =20).Combined responses reflected a range of none to little impairment (always or often able to perform task), to large impairment (sometimes, rarely, or never able to perform task).In all statements, the presence of pre-diagnosis pain had a greater frequency of large impairment.A greater percentage of those reporting the absence of pre-diagnosis PC pain were significantly better able to do chores such as vacuuming or yard work compared to those reporting its presence (P = .0025),and similar trends were observed with other measures of physical functioning (go up and down stairs at a normal pace, walk for 15 min, run errands).Patients with pre-diagnosis PC pain also reported Large impairment due Most patients reported discussing pain with their HCP, who recommended/prescribed medications for pain control (prescribed, over the counter analgesics and nonpharmacologic therapies), with a slightly higher frequency in those reporting the presence vs. absence of pre-diagnosis PC pain (98.6% vs. 83.3%,respectively, P = .027).Those reporting the presence of pre-diagnosis PC pain comprised 67%-100% of recommended/prescribed pain medications queried.Frequencies of prescriptions/recommendations reported for opioids, nonopioids, and over the counter medications were the same for both groups.Responses to additional therapeutic nonpharmacologic options were also queried.Responses of the presence of pre-diagnosis PC pain who provided NRS pain scores found highest frequencies of pain located in the The number (N) and frequency of responses (%) to items of the health assessment survey (column 1) from the PC patients is shown based on the presence (column 2) vs. absence (column 3) of pre-diagnosis PC pain.The total numbers of responses to the items are displayed in column 4. Percentages represent column percentages for non-missing data related to each item or question.For the pain intensity item, the NRS pain score responses from N = 155 responders were combined into 2 groups, low NRS 0-3 pain scores and high NRS 4-8 pain scores.For the remaining items, 2 score combinations are shown representing little impairment on statement activities (the top descriptors), or large impairment on statement activities (the bottom descriptors).The P-values calculated from the patient response differences from the large impact scores, are displayed (column 5).Bold values are statistically significant.

Effects of Pain Intensity and PC Experience
Of N = 1978 PC patients completing the immediately accessible patient information and PC experience surveys, N = 155 answered the NRS pain rating question in the health assessment survey as depicted in Supplementary Table S2.PC patient pain rating responses were within the numeric range of 0-8 out of a possible rating scale from 0 to 10, with high NRS 4-8 pain scores reported by 27.1% and low NRS 0-3 by 72.9% of respondents.The frequency of NRS pain scores = 0, or patients reporting no pain, was 36.8%, with 36.1% reporting NRS pain scores of 1-3.Both presence vs. absence of pre-diagnosis PC pain groups contained responses of NRS pain = 0 at a frequency of 31.2% vs. 50%, respectively.
Responses from patient information and PC experience surveys by PC patients assigned to high (NRS 4-8, N = 42) vs. low (NRS 0-3, N = 113) pain score groups, reported post-diagnosis at survey completion, are compared in Supplementary Table S3.The high vs. low NRS pain group had significantly lower average age at survey completion, was more likely to indicate lymph node metastasis, and reported decreased frequencies of no evidence of disease and having received surgery for PC treatment.The high-pain group had significantly increased frequencies of patients reporting pre-diagnosis abdominal and/or back pain, digestive problems, and fatigue.There were no differences in post-diagnosis symptoms between the high-and low-pain groups.There was also no difference in the percent of individuals with adenocarcinoma vs. neuroendocrine types of pancreatic cancer who reported pre-diagnosis pain, or reported high vs low NRS pain scores, despite the significant differences in the survival rates between the 2 pancreatic cancer histologies (data not shown).
To determine if the delay in survey completion (an average of 2 years after diagnosis) resulted in bias in recall of pain intensity, the percent of individuals reporting high vs. low pain and completing the survey within 2 months vs. greater than 18 months after diagnosis was compared (Supplementary Table S4).Although there is a trend for a somewhat higher percentage of individuals who reported high vs. low pain within those who filled out the survey shortly after diagnosis,  The Oncologist, 2023, Vol. 28, No. 12 e1193 this was not statistically significant and suggests there is no bias related to the time after diagnosis of survey completion.The responses of PC patients assigned to high vs. low pain groups in the health assessment survey were compared for their impairment in physical functioning, pain interference with normal activities, anxiety, fatigue, and depression (Table 4).Average ± SD of NRS pain scores between high and low NRS pain groups were 6.28 ± 1.05 vs.1.16± 1.39, respectively, P = .0001.High NRS pain scores associated with significantly increased frequencies of large impairment: of physical functioning, (4/4 questions, P = .001);pain interference, (4/4 questions, P = .0001);feelings of anxiety, (4/4 statements, P = .03-.003); activities related to fatigue, (4/4 statements, P = .001)and feelings of depression, (felt depressed, P = .02,with the remaining 3 statements trending in the same direction).

Frequency of Reported Pain Scores by Year of PC Diagnosis
To determine if trends in PC pain changed over the past decade, Table 5 shows high vs. low NRS pain scores based on the year of PC diagnosis.Of patients diagnosed in years prior to 2018 (N = 30), 23.8% reported high NRS pain scores at survey completion.Patients diagnosed with PC in years.2018-2020 (N= 1 2), 41.4% reported high NRS pain scores, P = .055,notably not significantly reduced over the 11-year span of PC diagnoses.Patients diagnosed with PC prior to 2018 vs. 2018-2020 had decreased frequencies of did not receive any treatment and increased frequencies of chemotherapy, radiation, and surgery as treatments, shown in Supplementary Table S5.

Discussion
This cross-sectional study presented unique single-time point responses to an online survey over a 5-year period between 2016 and 2020 from close to 2000 PC patients visiting PanCAN's Patient Services and voluntarily completing the Patient Registry survey.The most frequently reported pre-diagnosis symptom was PC pain, which aligned with previous reports. 2,26,27The initial assessment comparing the presence of pre-diagnosis PC pain showed significantly increased responses to symptoms reflecting GI dysfunction and GI-based local and distant tumor spread.9][30][31] Systemically, significantly increased associations of pre-diagnosis PC pain and impairment of functioning and mood were reported, consistent with previous reports. 7,9,32he second assessment of NRS pain scores found substantial impairment of daily physical activities, mood statements, and increased frequencies of pre-diagnosis PC pain with high (vs.low) NRS pain scores.Previous reports of poor outcomes have been associated with increased pain levels, depression and fatigue, decreased physical and social functioning, 5,7,32,33 and increased anxiety. 11,12he third comparison of resource utilization between the presence vs. absence of pre-diagnosis PC pain found increased ER and hospital visits and increased HCP recommendations for PC pain management.Increased frequencies of HCPs recommending/prescribing analgesics, including opioids, for most of the reported PC pain were noted, indicating HCP responsiveness to PC patient pain complaints.
Two time periods queried (pre-and post-diagnosis symptoms) of the PC patients' clinical courses showed that PC pain levels can be dynamic and most likely responsive to intervention.Some reporting the presence of pre-diagnosis PC pain also reported post-diagnosis NRS pain scores = 0 and alternatively, some patients reporting the absence of pre-diagnosis PC pain noted post-diagnosis NRS pain scores >0.NRS = 0 pain scores, or no pain, were reported in 37% of responders at post-diagnosis survey completion, in agreement with a previous study. 34C pain is reported with poor outcomes, as its onset and severe intensity are reported in advanced disease. 6The presence of pre-diagnosis PC pain was associated with increased frequencies of GI symptoms, all reported cancer stages (highest in borderline resectable), and spread to GI tissues.It is also associated with increased frequencies of higher NRS pain scores, large impairment of daily and social activities, and increased resource utilization in ongoing PC patient management.The presence of pre-diagnosis PC pain and high NRS pain scores appear to act as sufficient and also possibly additive or synergistic contributors to reported large impairment of activities and mood.The absence of pre-diagnosis PC pain was associated with significantly increased frequencies of jaundice, symptoms related to biliary obstruction, and earlier PC diagnosis. 9,14his study had several limitations.This was a cross-sectional study and self-administered, limiting patient responses to one-time point, post-diagnosis, with biases intrinsic to voluntary survey completion.Large numbers of responses obtained from the immediately available patient information and PC experience surveys were reduced to 5%-9% in the drop-down surveys of health assessment and pain management, precluding further analyses with associated variables and raising concerns of responder bias.Small patient participation in survey completion has been reported for voluntary surveys . 35Also, response interpretation of the pain-rating item was limited, as patient orientation to pain type, anchoring, location, time frame, etc. were unavailable.
These survey responses of pre-diagnosis PC pain signals strongly support a multitude of previous article positions on the need for effective, ongoing symptom and resource management, regardless of PC prognosis.It is notable that the frequencies of high NRS pain scores did not decrease over a recent 11-year span available for study, possibly due to diagnosis and treatment confounders, or at least in part due to undertreated pain.8][49] Their earlier involvement in patient care may require corporate or government policy changes to provide consistent evolving management of symptoms to improve HRQoL and overall health and survival in PC. 10,45,[50][51][52][53] Finally, the use of patient surveys querying symptoms and treatments independent of

Figure 1 .
Figure 1.Inclusion and exclusion criteria.The inclusion/exclusion criteria used for this study in the 58 months between 2016 and 2020 is depicted.PC patients who had no responses to the patient information or answered less than 70% of the items in the PC experience surveys were excluded from the study (N = 146, 6.5%, top, level 1, right side).Patients who responded "Yes" or "No" to the survey item "Before my diagnosis I had: abdominal and/or back pain, " (pre-diagnosis abdominal/back pain, N = 1978, mid-level 3, left side) were included.N = number of participants in each population, n = number of participants in subcategory.

aA
/B pain, abdominal and/or back pain.b Fisher's Exact test.
Other pain treatment modalities recommended/prescribed by my HCP, (N = 14 modalities) 134 28 NA ND Patient numbers (N) and frequencies (%) who indicated they are experiencing or have experienced pain related to pancreatic cancer from the pain management survey (column 1), based on PC patients reporting the presence (column 2) or absence (column 3) of pre-diagnosis PC pain.The total numbers of responses for each item are shown (column 4).Percentages values represent column percentages for non-missing data related to each item/question.The P-values for each item comparison are shown (column 5).Other pain treatment modalities recommended/prescribed by my HCP (N = 14): Herbal remedies, hot or cold packs, exercise, changing positions, (such as laying down or elevating your legs), physical therapy, massage, acupuncture, rest, guided imagery, relaxation techniques (hypnosis, biofeedback, and meditation), creative techniques, (art or music therapy), chiropractic treatment, osteopathic treatment, another recommendation not listed above.Combined number of responses from patients reporting the presence vs. absence of pre-diagnosis PC pain regarding the 14 additional treatment options were N = 134 vs. N = 28, respectively (not shown).Bold values are statistically significant.a A/B pain, abdominal and/or back pain.b Fisher's Exact test.

e1194
The Oncologist, 2023, Vol.28, No. 12 numbers (N) and frequencies (%) of high (column 2) and low (column 3) NRS pain score based on year of PC diagnosis divided into groups of diagnosis prior to 2018 (2010-2017) and diagnosis 2018-2020 (column 1).Percentages represent column percentages for non-missing data related to each item/question.The total number of responses for the years of PC diagnosis are shown in column 4 and the P-value obtained by comparing the high pain scores is shown in column 5. e1196 The Oncologist, 2023, Vol. 28, No. 12 B.D.E.D., L.M.M. Final approval of manuscript: All authors.

Pre-diagnosis A/B pain a N = 1222 N (%) No pre-diagnosis A/B pain a N = 756 N (%) Total N = 1978 P-value
Patients reporting the presence vs. absence of pre-diagnosis PC pain reported significantly increased frequencies of large impairment in 45% of 20 health assessment activity statements of physical functioning, pain interference, anxiety, fatigue, and depression (Table2).Pain intensity was increased in those with pre-diagnosis PC pain, with high NRS pain scores significantly associated with the presence (vs.absence) of pre-diagnosis PC pain (34.4% vs. 16.1%,respectively, P = .012).

Survey items in patient information and PC experience Pre-diagnosis A/B pain a N = 1222 N (%) No pre-diagnosis A/B pain a N = 756 N (%) Total N = 1978 P-value
a A/B pain, abdominal and/or back pain.b Fisher's exact test.

Table 2 .
Effect of pre-diagnosis pain on overall health.Significantly increased frequencies of ER visits were reported in patients also reporting the presence vs. absence of pre-diagnosis PC pain, N = 41 vs. N = 4, respectively, P = .018.Only individuals reporting pre-diagnosis pain reported >2 ER visits.Hospitalizations reported for PC pain management (N = 59) showed a similar trend.Patients reporting the presence vs. absence of pre-diagnosis PC pain comprised the majority of hospitalizations, N = 56 (95%) vs. N = 3 (5%).
Effect of Pre-Diagnostic Pain on Pain ManagementOf N = 1978 PC patients completing the immediately accessible patient information and PC experience surveys, N = 93 answered the pain management survey as depicted in Table3.PC patients reporting the presence vs. absence of pre-diagnosis PC pain were asked if they are experiencing (or have experienced) pain related to PC to assess their pain management resource utilization, which included emergency room visits, hospitalizations, and HCP visits/contacts for PC pain management.The NRS pain scores from those reporting pre-diagnosis PC pain were significantly higher, (avg ± SD) 2.64 ± 2.56 vs. 1.56 ± 2.01, respectively, P = .0039.The total number of patients reporting ≥1 ER visit for pain management was N = 45, and a total number of ER visits N = 86.

Table 3 .
Effect of pre-diagnosis pain on pain management.

Table 4 .
14,54,55f pain intensity on overall health.The Oncologist, 2023, Vol. 28, No. 12 e1195 clinical studies offers adjunctive instruments to spot trends, consequences, contributors, and mitigators of symptom intensities.Future use of PanCAN survey responses could further assess the "real world" therapeutic potential of SOC, off-label, experimental, and integrative treatments, such as radiation therapy, chemotherapy, surgical excision or ablation, and interventional analgesic delivery.14,54,55ConclusionPCpain is a prominent PC symptom in all PC stages, dynamic in intensity and causation, often reported in the pre-diagnosis interval, and often undertreated.Pain is a modifiable risk factor for poor outcomes and worse prognosis.Treatment of downstream consequences of undermanaged PC pain is less desirable and inefficient.Its successful mitigation will require early dedicated, ongoing pain management.A better alignment of tumor reduction with patient symptom reduction will contribute to improved functioning, HRQoL, and overall survival.The responses to items of the Health health Assessment assessment survey (column 1) from the PC patients with high (column 2) or low (column 3) NRS pain scores are shown.The total number of responses to items is displayed in column 4. Percentages represent column percentages for non-missing data related to each item/question.Two score combinations are shown under each statement to assess pain NRS scores, either little impairment on statement activities ( top descriptors), or large impairment on statement activities (bottom descriptors).The P-values calculated from the patient response differences from the large impairment groups are displayed (column 5).Bold values are statistically significant.
a Fisher's Exact test.

Table 5 .
Pain scores by year of PC diagnosis.