Have we improved pain control in cancer patients ? A multicenter study of ambulatory and hospitalized cancer patients

Background: Pain in cancer patients is recognised as a major health problem yet few studies of both inpatient and outpatient populations have been carried out. Objective: To assess the frequency, type, and characteristics of pain in adult cancer patients, including both inpatients and outpatients. Design and setting: Cross-sectional study of 1,064 adult cancer patients (437 outpatients and 627 inpatients) from 44 hospitals and/or long-term-care centres in Catalonia, Spain. Cancer patients suffering from pain of any aetiology for ≥ 2 weeks and/or under analgesic treatment ≥ 2 weeks were enrolled. Measurements: Demographic and pain data were collected. The Spanish version of the Brief Pain Inventory was used to assess pain. Results: Pain frequency was 55.3%.Pain was less frequent in outpatients than inpatients (41.6% vs. 64.7%; p < 0.001), although median pain duration was longer in outpatients (20 vs. 6 weeks; p < 0.001). Pain was assessable in 333 patients, and intensity was similar in both outand in-patients ; however, outpatients reported less improvement, less pain interference with daily life, and less pain related to the cancer per se. In both groups, patients with multiple myeloma (73%), breast (65%), and lung cancer (61%) were most likely to report pain. Conclusions: Pain in cancer patients, both ambulatory and hospitalised, remains a challenge for health-care professionals, health administrators, and stakeholders. Our study reveals the high level of pain and distress that cancer patients continue to suffer, a problem that is particularly notable in outpatients due to the intensity and duration of the pain.


Introduction
Cancer patients suffer from a wide variety of different pain syndromes 1 , most of which are related to metastatic spread or health status. 2 Multiple concurrent factors can cause pain, leading to pain syndromes that are usually chronic and can last for months, years, or even for the patient's entire life..4 Pain in cancer patients has been recognised as a major health problem that impacts both quality of life and health care provision..9.10.11.12In Spain, although several epidemiological studies have been conducted to assess particular aspects of pain (e.g., breakthrough pain 13 or neuropathic pain 14 ), to our knowledge, no comprehensive pain studies (whether national or regional) have been carried out in a cancer patient population.Moreover, data is lacking on differences in pain between adult cancer outpatients and inpatients.This knowledge gap, together with the need to better understand pain in cancer patients in our context, led us to design and carry out the comprehensive epidemiological pain study presented here.The aim of the study was to assess the frequency, type, and characteristics of pain in adult cancer patients, including both inpatients and outpatients.

Material and Methods
This cross-sectional study enrolled a sample of ambulatory and hospitalised cancer patients ≥18 years of age with a histological diagnosis of solid or haematological neoplasms.All recruited patients were in treatment at the oncology and/or palliative care (PC) service of acute care hospitals in Catalonia, or at a long-term-care (LTC) centre with dedicated PC resources, all of which are form part of the public, universallyaccessible National Health System of Catalonia, Spain.
To obtain a representative sample of patients, we estimated that patients from 44 hospitals and LTCs were needed.According to the Catalonian Healthcare Services 2002 Directory, these 44 centres provided coverage for >80% of cancer patients in the region.
Assuming a cancer pain frequency of 50%, with a 5% absolute precision and 95% confidence interval (95%CI), statistical calculations indicated that a minimum of 624 inpatients and 435 outpatients would be needed to estimate the pain frequency in both populations.Since outpatient clinics are mostly tumour-type specific, the outpatient sample was restricted to the eight most common neoplasms (colorectal, lung, breast, prostate, bladder, leukaemia, non-Hodgkin's lymphoma, and multiple myeloma) to avoid potential bias in pain frequency related to over-or under-representation of particular tumour types.The number of cases needed for each of these tumour types was calculated according to the incidence rate in our region 15 .Hence, each hospital (based on the estimated tumour incidence) was expected to enrol a specific number of patients of each tumour type.Finally, patients were randomised for enrolment according to the weekly list of patient consultations.
The study was approved by the ethics committees of all participating centres.All patients signed informed consent forms.The participating hospitals and the local investigators are listed in the Appendix.
Patients were screened between January 2009 and December 2010, inclusive.
The clinical records of a randomized group of patients were reviewed by a specially-trained health-care professional interviewer (doctor or nurse)-assisted, in most cases, by the patient's treating physician-to identify the patients with pain.
Patients were considered to have pain if they reported any pain (regardless of aetiology) and/or the medical records showed that the patient was under analgesic treatment lasting ≥2 weeks.Patients with pain were invited to participate and asked to sign the informed consent form.

Cognitive status was checked by the Spanish version of the Mini-Mental Status
Examination (MMSE) before the pain assessment 16 .Patients with cognitive failure were excluded from further assessment in this study.
Given the inherent differences between inpatients and outpatients (different patient profiles, treatment regimens, and schedules), two distinct groups were considered: inpatient sampling was performed randomly whereas outpatients were stratified by tumour type.Demographic, clinical and pain variables collected are shown in Table 1.

Statistical analyses:
Categorical variables are shown as percentages with 95% CIs.Continuous variables are presented as means and standard deviations (SD) or as medians and interquartile range.Categorical data were compared using Pearson's χ 2 and Fisher's exact test, while the linear-by-linear χ 2 test was used for trends.Continuous variables were compared using the Student's t-test or Mann-Whitney tests depending on the distribution (normal or non-normal).A value of p<0.05 was considered statistically significant.All analyses were performed with the SPSS package (v.13 for Windows).

Results
Of the 1064 patients initially screened, 588 (55.3%) met the study criteria for pain, with a significantly lower percentage of outpatients vs. inpatients meeting these criteria(41.6%vs. 64.7%;p < 0.001).Of the 588 patients, 373 provided informed consent and 333 of these were cognitively intact and, therefore, eligible for pain assessment (Figure 1). in both solid and haematological malignancies.We observed no significant association between pain frequency and time-elapsed since cancer diagnosis.Pain was more common in patients being treated with radiotherapy and less frequent in those receiving chemotherapy.The tumour types (in both inpatients and outpatients) with the highest percentage of patients with pain were multiple myeloma (73%), breast cancer (65%), and lung cancer (61%).In contrast, tumours with the lowest pain frequency were lymphoma (40%) and leukaemia (25%) (data not shown).

As shown in
When comparing the 333 patients assessable for pain by place of assessment (Table 3), we found that ambulatory patients had a significantly better performance status, less advanced disease, longer time with cancer, and longer duration of pain than inpatients.Notwithstanding these findings, no significant differences in pain intensity, assessed according to average and worst pain, were observed.
Mean pain intensity and interference are shown in Table 4.Even though mean pain interference was lower in outpatients than inpatients, there were no significant between-group differences in pain intensity.On the whole, men experienced more pain interference in their 'relations with others' and more pain interference with 'sleep' than women (data not shown).No differences were found on other pain dimensions.Patients with a KPS ≤50 reported significantly more pain interference than those with KPS >50.
Patients with metastatic disease also presented significantly more pain interference than those without metastasis.All Brief Pain Inventory (BPI) interference dimensions were more significantly affected in patients with KPS ≤50, and in patients with metastases.
Table 5 shows the main pain types (and their pathophysiology).No statistically significant between-group differences (gender, age, KPS, time-from-cancer diagnosis and cancer stage) were observed in terms of the pain types and probable pathophysiological mechanisms.
The percentage of patients with tumour-related pain was significantly higher in the inpatient group vs. outpatients (81.3% vs. 50.4%;p < 0.001) (data not shown).
Conversely, outpatients reported higher levels of pain unrelated to the tumour or the anti-tumour treatment.Moreover, in patients with metastatic disease, tumour-related pain was significantly greater than in patients with no evidence of cancer or locoregional spread of the disease; 80% vs. 49.6%p < 0.001, respectively.

Discussion
The present study evaluated the largest sample to date of an unselected population of both ambulatory and hospitalised cancer patients in Southern Europe suffering from pain.To our knowledge, it is the first such study carried out in Spain.
The findings confirm that the pain frequency in our region (55.3%) is in-line with rates reported in other countries 10,22,23,24,25,26 .This result is also consistent with those from a large meta-analysis of 52 studies, which reported an overall pain prevalence of 53% 24 .
In addition, our data agree with previous reports indicating that pain frequency was higher in patients with poor functional status 10 and in patient with advanced disease.
Taken together, the data suggest that, on average, half of all cancer patients will experience pain.Some authors 27 have suggested that the percentage of patients reporting severe pain (NRS ≥7) in the 'average' and 'worst pain' categories could be used as an indirect indicator of poor pain control.In our sample, these percentages were 13.2% and 61.6%, respectively, without significant differences between in-and out-patients.Caraceni et al 2 reported similar findings.In their study, those authors found 20% had severe 'average pain' while 67% presented severe 'worst pain'.It appears that these percentages have not changed substantially in the time that has passed since Caraceni et al. published their report, eleven years ago.Consequently, this suggests that pain in cancer patients may not have improved as much in Europe and in Spain as some authors have claimed. 5,28 our study, the prevalence of moderate-severe pain (NRS >5) was 31.5% which is in agreement with authors who have reported a range from 20-45%. 22,23,24,29,30 most studies, the mean pain duration ranges from 6 months to 1 year. 2,18 erestingly, inpatients had a shorter median pain duration (6 weeks) than outpatients (20 weeks); however, the period of pain was protracted in both groups, and this is indicative of the amount of pain that patients in our study sample had to tolerate.
With regard to patients' reported pain relief, the mean improvement was slightly more than 61%, similar to the pain improvement (65%-68%) found in other studies. 3,31in interference in our sample was similar to the rates reported by Daut 3 and Strassels. 32Of note is that both inpatients and outpatients in our study reported similar rates of mean 'average' pain; however, outpatients reported significantly less pain interference and pain relief.It is not clear if this finding represents an adverse impact (or dissatisfaction) associated with extended periods of pain, although pain intensity has been recognised as the major predictor of pain interference with daily-life activities 3 .
Nevertheless, it is well-known that the reported intensity varies in response to factors such as the meaning of pain, 3,33 breakthrough pain, 34 psychological distress, 35 and pain education. 36Although several studies have described multiple reasons underlying inadequate cancer pain management in outpatients, 37,38,39 those same studies have also indicated that this is a complex, multi-factorial problem.In the present study, we too found it difficult to pinpoint the reasons for poor pain alleviation in our outpatient group.For this reason, it is evident that further research into this complex topic is warranted; however, it is important physicians, nurses, and health authorities be aware of the existence of this issue.
We found that, compared to men, a significantly higher proportion of women reported pain.A recent review concluded that women appear to have lower pain thresholds, greater ability to discriminate painful sensations, higher pain ratings, and lower tolerance of pain. 40As in other studies, we did not observe differences in pain prevalence in relation to age. 22,23,24 our study, the proportion of patients on chemotherapy who reported pain was lower than those not receiving chemotherapy, regardless of cancer stage.This finding should not be interpreted as cause and effect, since many confounders can be involved and because this was an incidental finding of the present study, it is difficult to determine the reasons for this difference.To date, very few studies have assessed the potential analgesic role of chemotherapy, and further research is needed. 41,42,43It is well-known that pain prevalence is higher in patients receiving radiotherapy vs. those without radiotherapy 43 and our findings confirm this.However, as other authors have noted, this higher prevalence may be due to the preference to use radiotherapy for pain palliation.
We found that nearly 70% of pain was directly related to the cancer itself.
Hospitalised patients were more affected by cancer-related pain than outpatients, a finding that is consistent with other reports. 10,31However, these findings must be contextualised, as differences in patient characteristics between different studies can impact the results: in our study sample, only about 50% of patients had metastatic disease vs. 70% in the study by Caraceni et al 2 .
Anti-tumour treatment is another source of pain in cancer patients.Previous estimations have suggested that treatment-related pain accounts for 17-20% of pain 2,3,44 .We found, in contrast, that treatment-related pain accounted for only 6.6% of all pains in our sample.Conversely, for non-cancer related pain, we found that nearly 20% of all pains were due to this pain type, a finding that contradicts other studies 2,3 in which non-cancer related pain ranged from 2.3-9%.The reason for this divergence is not clear, although it could be due to the bias in better KPS in ambulatory patients (median KPS of 80%), 11.9% of whom were cancer-free.Nevertheless, outpatients in our sample experienced significantly more treatment-related pain (and the sequelae thereof) compared to inpatients, in addition to more pain unrelated to the disease or its treatment.However, it is important to note that non-cancer related pains can be just as intense as cancer-related pains and can have an equal or even greater interference on daily-life activities, depending on the nature of the pain. 3 our study, around 40% of patients experienced pain of mixed origin, a percentage that is similar to the 43% reported in other studies. 2 Neuropathic pain was less common in our study (5.2%) compared to the Caraceni et al 2 study (7.7%).If overall pain with neuropathic component is pooled, the frequency of neuropathic pain rises to 30% in our study, similar to the figure reported in a recent multi-centre study (33%). 14mitations worth noting.First, patients were considered to have prolonged pain-and this was an inclusion criterion-when the pain lasted for ≥ 2 weeks or the patient was under analgesic treatment ≥2 weeks.This cut-off point could be controversial because, according to the International Association for the Study of Pain (IASP) 45 , chronic pain in non-malignant conditions is defined as pain lasting 3 or more months.However, as that same report notes, a 3 month cut-off value is excessive in cancer patients, many of whom may have a life-expectancy that is shorter than this period.As a result, we believe the cut-off point should be adjusted by clinical experience.In our experience, a 2-week cut-off point allows us to rule out acute and transient pain conditions, and thus a pain duration ≥ 2 weeks seems appropriate to classify a pain as prolonged in this specific patient population.A second potential limitation of our study is the relatively low participation rate, which was approximately 60% for the inpatient group.This was due to the fact that some patients simply did not wish to participate.However, other similar multi-centre studies present similar participation rates ranging from 51% 5 to 66%. 22A third and final potential limitation is that our sample did not include all possible tumour types (notably head and neck cancers) in which pain is highly prevalent 46 .Neverthess, we did include the most common cancer types and although it would have been desirable to assess more cancers, doing so would have been impractical due to the large number of patients required and because the present study was not designed to assess the pain prevalence for each cancer type (even though the data obtained accurately characterises pain prevalence in the study population.

CONCLUSION
The present study adds substantial new and comprehensive epidemiological information about the pain suffered by cancer patients in our country, both in-and outpatients.Although the importance of cancer pain relief has become more widely recognized, cancer-pain remains a major problem, despite the considerable efforts made by health-care providers in recent decades to improve pain alleviation.
Our study reveals the high level of pain and distress that cancer patients in our region still suffer, a problem that is particularly concerning for outpatients due to the intensity and duration of the pain.More efforts need to be made to increase awareness of this problem and to find better solutions.Patients with chronic pain n=588

Patients cognitively intact: n=333
Intact cognition: Assessable for pain: Patients assessable for chronic pain: n=333

Table 2 ,
patients in LTCs reported significantly more pain than those in acute hospitals.Similarly, patients under PC care reported more pain than those in Medical/Radiation Oncology, and Haematology services.Pain was more frequent in women.No significant differences in pain were observed by age.Pain frequency increased significantly in relation to Karnofsky score (KPS) decline and cancer spread