Pain knowledge of patients and family caregivers as predictors of pain management outcomes in cancer patients: a multicenter study in China

The purposes of this study were to evaluate the current situation of pain management outcomes, patients’ pain knowledge, and family caregivers’ pain knowledge in China and explore the impact of pain knowledge of patients and family caregivers on pain management outcomes. A total of 410 patient-family caregiver dyads were assessed by Brief Pain Inventory (BPI), the Pain Knowledge subscales of Patient Pain Questionnaire (PPQ), and Family Pain Questionnaire (FPQ). The difference in pain management outcomes was analyzed using bivariate analysis first, and then variables with statistical significance in bivariate analysis were included in multivariable linear regression analysis. The average patient pain score in the last 24 h was 3.23 (SD = 2.16). The total average scores on pain knowledge of 410 patients and family caregivers were 3.60 ± 1.90 and 3.57 ± 1.76. Multivariable linear regression analysis showed taking strong opioids, patients’ perceived moderate health status, patients’ pain knowledge, and family caregivers’ pain knowledge were the main factors influencing the pain management outcomes. Pain knowledge of patients and family caregivers were important indicators of pain management outcomes, indicating tailored cancer pain education program should be developed.


Introduction
Pain is one of the most common symptoms suffered in cancer patients. Unrelieved pain greatly affects patients work, activities, motivation, mood, and overall quality of life, and denies patients comfort [1].
A systematic review of literature published between 1966 and 2005 showed that the overall prevalence of pain in cancer patients with all stages was 53%, and moderate to severe pain was reported by 31% of all patients [2]. In recent years, the whole world has made a lot of efforts in pain management. New insights in pain mechanisms and the development of new analgesics, in combination with the increased attention for cancer pain, gave hopeful reasons to expect a decrease in the prevalence and severity of cancer pain [3]. However, an update systematic review of literature published between 2005 and 2014 showed that the overall prevalence of pain in patients with all cancer stages was 51%, and moderate to severe pain was reported by 33% of all patients [4]. Therefore, it is suggested that the prevalence and severity of pain in cancer patients may have not decreased over the world.
However, there are great differences in cancer pain prevalence and severity among different countries [5]. Therefore, it is more significant to explore the different cancer pain management situations and influencing factors in various countries or regions for the improvement of pain management. In China, the progress in government policy, public advocacy, medication availability, and education for health providers, patients, and family caregivers have changed the current pain relief situation in the last decades [6]. However, there are few multicenter studies that can fully understand the current situation of pain management in China.
Furthermore, inadequate pain management can be attributed to barriers related to patients, family caregivers, healthcare professionals, and the health-care system [7]. There are many factors involving patients influencing pain management, such as age, education, and other demographics [8]; cancer diagnosis, stages, and other disease-related factors [9]; as well as pain knowledge and belief [10][11][12]. Additionally, as health-care services growingly move off hospitals, the vast majority of cancer patients' care is more often provided with the help of family caregivers in the community over extended periods of time. The basic characteristics, cancer pain knowledge, and beliefs of family caregivers have great impact on patients' pain [13,14]. But the results of various studies on the influencing factors involving patients and caregivers are inconsistent [9,[15][16][17].
According to Knowledge, Attitude/Belief, and Practice (KAP) model [18], patients' pain knowledge has an impact on their beliefs, and then affects their pain management behavior, which will influence the outcomes of pain management [15]. In addition, it should be noted that family caregivers deliver complex cancer pain assessment, communicating with medical stuff, administering analgesics, and managing side effects. Therefore, it is hypothesized that the pain knowledge of family members will also have an impact on the pain outcomes of patients. However, few studies have explored the impact of pain knowledge of patients and their families on pain management outcomes [19].
In view of the current situation of pain management, the influencing factors especially the pain knowledge of patients and their families are not clear. The aims of our study were to (1) evaluate the current situation of pain management outcomes, patients' pain knowledge, and family caregivers' pain knowledge in China and (2) explore the pain knowledge of patients and family caregivers on pain management outcomes.

Patient population and study design
A multicenter descriptive cross-sectional study was conducted between June and December 2019. A convenience sample of 444 dyads cancer inpatients and family caregivers was recruited from seven hospitals in China. Of the 444 dyads, 410 dyads answered all our questions and completed all the questionnaires.
Patients were eligible to participate if they were over the age of 18; have been given a diagnosis of cancer; able to communicate in Chinese; and gave informed consent. Patients were excluded if they were too weak to complete the questionnaire. The family caregivers had to be involved in patients' care, which were identified by themselves or patients, over 18 years old, and able to communicate in Chinese. The sample content should be at least 5-10 times the number of independent variables according to the requirements of multifactor analysis for the sample [20]. The number of independent variables involved in this study was 35; as such, 175-350 patient-family caregiver dyads were needed. Considering a 90.0% effective response rate, the sample size was at least 195 dyads.
This study was approved by the Ethics Committee of Peking University Cancer Hospital. A training session was scheduled for the research nurses from seven centers to establish data collection procedures and uniformity. Eligible patients were approached by a research staff member. Written informed consent was obtained from all patients and family caregivers.
We reported our results conforming to the Strengthening the Report of Observational Studies in Epidemiology (STROBE).

Pain management outcomes
The current pain management situation was evaluated by the Brief Pain Inventory (BPI). BPI was developed by Cleeland in 1989 for use with cancer patients for the assessment of epidemiology, clinical pain, and the effectiveness of pain treatment [21,22]. The BPI measures both the intensity of pain (pain now, worst, least, and average pain in the last 24 h) and the interference of pain with daily functioning. For item rating, 0-10 numeric scales were used with 0 being "no pain" and 10 being "pain as bad as you can imagine." The BPI asks for ratings of the degree to which pain interferes with general activity, mood, walking, normal work, relations with others, sleep, and enjoyment of life using numeric scales, with 0 being "no interference" and 10 being "interferes completely." BPI also queries the patient about pain relief rate in the last 24 h, with 0 being "no relief" and 100% being "complete relief." Xin Shelley Wang [23] translated and adapted BPI into Chinese, the Chinese version of the Brief Pain Inventory (BPI-C). The coefficient alphas for the pain intensity subscale and pain interference subscale were 0.849 and 0.915, respectively, and the constructive validity, convergent validity, and inter-rater reliability of BPI-C were also well documented.

Pain knowledge
Patient pain knowledge Patient pain knowledge was evaluated by the Pain Knowledge subscale of Patient Pain Questionnaire (PPQ). The PPQ has been used to assess cancer patients' pain knowledge and experience since 1991 [24]. In 2000, Betty Ferrell published an open access updated 16-item edition on the website of the City of Hope National Medical Center [25] in the USA, including 9 items on the knowledge subscale and 7 items on the experience subscale. The two subscales can be used separately. Pain knowledge subscale included 9 items, which were fully presented in Table 4. The Likert 11 score (0-10 points) was used. The total subscale score (0-10) was the average score added by the scores of each item. The higher the score, the worse the knowledge. A series of psychometric analyses were performed on the PPQ instrument including content validity (CVI = 0.95), test-retest reliability (r = 0.65), internal consistency (alpha = 0.74), and factor analysis established with caregivers (N = 219) [25].
Family caregiver pain knowledge Family caregiver pain knowledge was evaluated by the Pain Knowledge subscale of Family Pain Questionnaire (FPQ). The FPQ was originally developed in 1989 and was revised by Betty Ferrell in 1993 [26]. In 2000, Betty Ferrell published an updated 16-item edition on the website of the City of Hope National Medical Center [25] together with the PPQ. The items (Table 5), scoring method, and the total score of pain knowledge subscale were identical to PPQ, and the difference between these two scales was language expression. A series of psychometric analyses were performed on the instrument including content validity (CVI = 0.90), concurrent validity (r = 0.60, p < 0.05), factor analysis, and test-retest reliability (r = 0.80) established with a retest of caregivers (N = 67) [25].
Brislin's translation method proposed in 1970 [27] was used for forward-and back-translation of PPQ and FPQ. Initial translation was carried out by two independent translators. Two results of the translations were then synthesized and discussed within the research group, consisted of one researcher and two translators, until the consensus was reached. Two authorized Chinese English bilingual nursing expert who were blind to the original English version performed back-translation independently. Two back-translated documents were compared with the original to make a consensus draft. Then the final version was developed. All items and subscales were consistent with the original scale. We tested the internal consistency reliability in our study. Cronbach's alphas were 0.701 and 0.620 for the Chinese Pain Knowledge subscale of PPQ and FPQ, respectively.

Basic characters
Basic characters of patients included hospital, gender, age, education, employment, locality, self-perceived health status, cancer diagnosis, anti-tumor therapy, analgesics, and analgesic adherence. Administrative districts of China include prefecture-level city, county-level city, town, and countryside [28]. So the locality was classified into city, town, and countryside. Self-perceived health status was assessed by a question "How are you feeling about your health status?", and patients classified their status as good, moderate, or bad. Analgesic adherence was identified if patients regularly took all analgesics following prescriptions. Analgesic prescriptions were collected from medical records and how patients took them was reported by their own selfreports. Basic characters of family caregivers included gender, age, education, employment, relationship with patients, and living with patients or not.

Statistical analysis
Data were analyzed using the SPSS version 20.0 (IBM Corp., NY, USA). Descriptive statistics and frequency distributions were generated on basic characteristics of patients and family caregivers. The scores on patients' pain knowledge and family caregivers' pain knowledge were reported with mean and standard deviation. Indicators for pain management outcomes were analyzed using the two independent samples t test, one-way ANOVA analysis, and Wilcoxon rank sum test for bivariate analysis. Variables with statistical significance in bivariate analysis were included in multivariable linear regression analysis. Stepwise regression analysis was used, α in = 0.05, α out = 0.10, and P < 0.05 was taken as the significant difference standard. Ordinal categorical variables were regarded as continuous variables for statistical analysis, and dummy variables were used for analysis of unordered categorical variables. And final two-sided P values < 0.05 were considered statistically significant.

Characteristics of patients
Four hundred ten patient-family caregiver dyads participated in our study. One hundred twenty-two (29.8%) patients were recruited from cancer hospitals, and 288 (70.2%) were from general hospitals. The mean (SD) age of the all patients was 59.34 (12.59). The majority of the patients were male (57.1%). Most of the patients perceived moderate health status (60.7%), were educated in junior high school or below (67.1%), and were unemployed (85.1%). In our study, nearly half (44.9%) of the patients lived in cities. The most diagnosis (32.9%) was lung cancer. Strong opioids has been prescribed for 72.4% of the patients, and 82.7% of the patients completely complied with orders ( Table 1).

Characteristics of family caregivers
The mean (SD) age of 410 family caregivers was 48.25 (13.27). Nearly half of participants were female (54.9%),

Bivariate analysis of variables and patients' average pain score in the last 24 h
The average pain severity in the last 24 h was the main pain management outcome of our study. Pearson correlation analysis was used to analyze the correlation between average pain severity and patients' age. The result showed that the correlation coefficient was 0.046 (P > 0.05). Two independent samples t test, ANOVA, Wilcoxon rank sum  test were used to analyze the correlation between average pain severity and patients' basic characteristics. The results showed that the average pain intensity was correlated with patients' locality, self-perceived health status, anti-tumor therapy, and analgesics (P < 0.05) ( Table 1). The results of two independent samples t test, ANOVA, and Wilcoxon rank sum test showed that patients' average pain intensity was correlated with the relationship between family members and patients, whether family members and patients lived together or not (P < 0.05).
The results of Pearson correlation analysis showed that 6 out of 9 patient pain knowledge items were correlated with patients' average pain intensity (P < 0.05) (Table 4). Additionally, 8 out of 9 family caregivers pain knowledge items were correlated with patients' average pain intensity (P < 0.05) ( Table 5).

Predictors of patients' average pain severity
Based on the results of univariate analysis, the basic characters of patients, demographics of family caregivers, pain knowledge of patients, and pain knowledge of family caregivers significantly correlated with patients' average pain severity were included in multivariable linear regression analysis. The results of multivariate analysis showed that R 2 = 24.4%, adjusted R 2 = 23.3%, F = 21.661, P < 0.001, and D-W = 0.550 (Table 6).

Discussion
Cancer pain is one of the most common, debilitating symptoms among cancer patients worldwide. In our study, the average pain score in the last 24 h was 3.23 (SD = 2.16), suggesting moderate pain intensity. Of patients, 29.3% and 8.0% reported moderate and severe pain; also, the mean pain relief rate in the past 24 h was only 66.46% (SD = 27.13). The pain management outcomes were not satisfied. During the past few decades, lots of efforts have been made to improve the cancer pain management in China, involving government policy and approach, medication availability, education for health providers, patient education, and public advocacy [6]. However, the results suggest that we still have a long way to go for managing cancer pain.
Patients in this study reported a similar level of pain knowledge (3.60 ± 1.90) with family caregivers (3.57 ± 1.76). What is interesting is that the three items with the worst pain knowledge of caregivers were consistent with those of patients. The three items were "prognosis," "drug tolerance," and "addiction risk." A previous systematic review indicated that attitudinal barriers to cancer pain management across patients, family caregivers, and the general public were similar, which supports our finding [29]. A study in 2004 in mainland China assessing the cancer pain beliefs also found that concerning  about cancer prognosis, tolerance and addiction were patients' and family caregivers' main belief barriers [30].
There have been many cancer pain educational programs in China for both patients and families, and educational interventions have made significant benefits of patients on knowledge and attitudes [6]. However, due to the increasing number of cancer patients, cancer pain education programs may not meet all the knowledge needs of cancer patients. Therefore, the main concerns in about 20 years ago in China were still concerns in patients recruited in present study, despite lots of efforts. The results suggest that in order to improve the knowledge of patients and their families, we should strengthen cancer pain education from the following three aspects: provide standardized education for more cancer patients, families, and the public through the network; take more targeted measures, such as individualized education based on cancer pain knowledge assessment; and take more diversified educational measures, such as culturally tailored video, booklet, telephone, mini fellowship, and face-to-face consulting [31]. In addition, it is also very important to improve the ability of pain education of medical staff.
The regression model showed that the use of strong opioids, patients' self-perceived health status, patient's knowledge (too much analgesics, addiction), and family caregivers' knowledge (efficacy of cancer pain management, too much analgesics) were predictors of pain management outcomes. Firstly, the results indicated that the use of strong opioids had the greatest impact on the outcome of pain management. The pain intensity of the patients using strong opioids was 1.172 times higher than that of other patients. This result indicates that the pain is still poorly controlled despite the use of strong opioids. Why did this happen? Physicians may not change previously prescribed analgesics based on patients' current pain status, and another study also supported this idea [32]. The results indicate that the standardization of pain treatment still needs to be strengthened. Pain reassessment should be performed at specified intervals to ensure that analgesic therapy is providing maximum benefit with minimal adverse effects [1]. Also, strong opioids are not effective enough for some of patients with neuropathic pain or refractory pain; if necessary, referral to a pain specialist and/or the use of interventional strategies should be considered [1].
Secondly, patients self-perceived health status was one of the influencing factors of pain intensity in our study. Patient-perceived health status is a comprehensive perception of physical, psychological, mental, and other aspects of the state. So we used a question to assess this item. Pain refers to a subjective experience, and Dame Cicely Saunders, the founder of the modern hospice movement, applied the term total pain as having physical, psychological, social, and spiritual components interacting upon one another [33]. Therefore, the comprehensive health perception of patients is related to the pain experience of patients.
Thirdly, what consistent with our hypothesis was that patients' pain knowledge was a predictor of the pain management outcome, and two knowledge variables were significant. Of the variance, 6.7% was due to one patients' knowledge "believing that patients are often given too much pain medicine." Patients who held this specific knowledge reported more pain. Another factor that affects pain management outcomes is addiction. Early work by Xiaoxiao Ma [34] in 146 Chinese cancer pain patients found that tolerance, addiction, and side effects were main belief barriers. We consider that because of these concerns and worries, patients hold the view that they were often given too much pain medicine, which leads to the behavior of not reporting pain, refusing to add dosage, not taking medicine on time, and stopping taking medicine without orders, which lead to poor pain control. Another study recruiting 122 cancer pain patients found that patients who knew that an increased need for analgesics reflected a real increase in their pain reported less pain, compared to those who did not have that knowledge [35]. This can also help explain the patient's perception that they are given too many pain medicines, leading to the above nonadherence behaviors. Thus, it can be seen that improving knowledge about the nature of cancer pain and its management is crucial in clinical pain management.
Finally, in the present study, pain management outcomes reported by patients were associated only a small way with family caregivers' knowledge of cancer pain and its management. Even though the family caregivers' pain knowledge on "efficacy of cancer pain management" and "patients are often given too much pain medicine" influenced patients' pain severity, the effect was small. A systematic review published in 2019 provides strong evidence that lack of pain knowledge among cancer patients, family caregivers, professionals, and the public was reported as one of the most common barriers to effective cancer pain management. This comprehensive systematic review included 36 studies about cancer pain knowledge or belief from 18 countries, but only 4 of them were conducted in family caregivers [29]. So more updated studies are needed to generate more contemporary data regarding to family caregivers. Further investigation is needed to determine how family caregivers' pain knowledge plays a role in patient cancer pain management.
Our study also has some limitations. First, convenience sampling method was used to recruit medical centers and patients. Even though the sample size was sufficient enough, the sample representativeness may be not very good. Second, family caregivers but not primary family caregivers were investigated in our study. Primary family caregivers involves more than general family caregivers in delivering care and support services to patients with cancer pain. Therefore, more attention should be paid to the primary caregivers in the future studies.
In summary, this study concludes that patients' pain management outcomes, patients' pain knowledge, and family caregivers' pain knowledge are not satisfactory. Moreover, this study reveals that pain management outcomes are obviously influenced by analgesics, self-perceived health status, and pain knowledge of patients and family caregivers. Advancing pain management by standardizing and individualizing pain treatment and implementing tailored education program to patients and family caregivers are important to improve cancer pain management outcomes.