The Effect of Chronic Pain on Diabetes Patients’ Self-Management

Data were collected as part of a larger multisite diabetes quality of care study (20,21). Human studies committees at participating Department of Veterans Affairs (VA) medical centers approved the study protocol. A full description of the sample criteria, response rate, and respondent characteristics for this larger study are described elsewhere (21). Briefly, potential study participants included patients with at least one VA nonancillary care outpatient visit between 1 October 1998 and 30 September 1999 at one of five regional VA healthcare systems and who were identified as having diabetes using data (pharmacy, laboratory, and diagnosis) obtained from the automated VA patient record system. Patients who met the initial selection criteria were invited to participate in a baseline computer-assisted telephone interview after verifying that they 1) had diabetes and 2) received most of their diabetes care through one of the participating facilities. We were unable to reach all individuals to verify their eligibility. However, assuming that persons we were unable to contact or could not confirm eligibility had the same rate of eligibility as those contacted and using a calculation endorsed by the Council of American Survey Research Organizations (22), the survey response rate for the overall study sample was 57%.

Data for this research were primarily obtained through a self-administered survey of study participants conducted ∼18 months after the initial phone interview. Of 1,359 study participants, 993 (73%) responded to this follow-up survey, which was conducted in March through May of 2003. Analyses of the baseline characteristics of those who responded to the follow-up survey versus those who did not show that respondents were slightly older, more likely to be white, more likely to have a high school or post-high school education, and had better reported health status at baseline.

The survey included several questions to elicit information about the extent to which patients experienced and were affected by chronic pain. Consistent with prior studies (6,9), patients with chronic pain were identified as those who reported having pain that was present most of the time for 6 months or more during the past year. Patients reporting chronic pain were then asked questions about pain severity and the impact of pain on behavior and moods (23). Information about pain location and treatment was also requested from those patients with chronic pain.

Diabetes self-management was assessed by asking patients to indicate their difficulty with and execution of key self-care practices as recommended by their doctor (24). These practices included taking diabetes medication, exercising regularly, following a recommended eating plan, checking their blood glucose level, and checking their feet for wounds or sores. Patients were asked to rate their level of difficulty with each activity on a scale ranging from 1, “so difficult that I couldn’t do it at all,” to 5, “not difficult I got it exactly right.” A “does not apply” option was also provided. These self-management items can be analyzed individually or combined to form a single scale ranging from 0 to 100 (higher scores indicate less difficulty with self-management). In a prior study, higher scores on this scale were found to be associated with better glycemic control and the receipt of recommended diabetes services (24).

Other factors that could affect self-management include general health status, the priority the patient gives to taking care of his or her diabetes, and the presence of depressive symptoms (8,25,26). Information about these factors was also collected as part of the survey. General health status was assessed using the general health question from the SF-36 questionnaire (27). Diabetes care priority was measured using a question that asks patients how much they agree (from 1 = strongly disagree to 6 = strongly agree) with the statement that taking care of their diabetes is not a high priority for them right now. Depressive symptoms were measured using the Center for Epidemiologic Studies Short Depression Scale (CES-D 10), in which a score of 10 or greater is considered a positive screen for probable depression (28).

Supplemental data, including sociodemographic information and BMI, were obtained as part of the baseline telephone interview, while information about other comorbid conditions was collected through a baseline medical chart review (21). A comorbidity measure was constructed by counting the number of other comorbid conditions from a list of 13 conditions (see Table 1 footnote).

Our primary outcome variables for all analyses included the overall self-management score as well as the five specific self-care items. We examined both the individual items as well as the overall score. Although we did not have specific hypotheses about how pain would affect an individual self-care item, we felt it was important to identify those activities that might be most affected by chronic pain. The self-management score was analyzed as a continuous measure, with a higher score representing better self-management, while the five self-care items were dichotomized, with respondents who indicated the activity was difficult (“but they managed some of the time”) to so difficult (“that they couldn’t do it at all”) coded as having difficulty with the activity or behavior.

The key independent variable for our first set of analyses was the self-reported presence of chronic pain. These analyses were conducted using the entire study sample except for 65 individuals who did not respond to the chronic pain question. A missing response to this item was primarily related to age and education level, with nonresponders being older and less likely to have a high school education compared with responders. Other variables of interest, which were also dichotomous, included the depression screen (i.e., CES-D 10 score ≥10), health status (poor or fair health versus good, very good, or excellent), the presence of other comorbid conditions (one versus none, two or more versus none), and whether diabetes was a patient priority.

The second set of analyses focused only on those patients with chronic pain and the association between pain severity and patient self-management. Severity was assessed using a question that asked about the general amount of bodily pain the patient experienced during the past 4 weeks. For these analyses, the response categories were collapsed and severity analyzed as a dichotomous variable (severe or very severe versus none to moderate). A variable indicating whether the patient used medication to control their pain on a regular or occasional basis was also added for this set of analyses.

We used bivariate analyses (χ² and Student’s t tests) to examine differences in the sociodemographic characteristics and health status of patients with chronic pain versus those without pain. Multivariable regression techniques were then used to examine the association between the presence of chronic pain and pain severity and patients’ reported difficulty with diabetes self-management while adjusting for other potentially influential factors. The overall self-management score was analyzed using ordinary least squares (OLS) regression, and difficulty with each of the individual components was analyzed using logistic regression. We report our results using the unstandardized coefficients from the OLS models and the adjusted ORs from the logistic models with their 95% CIs. The reported results were adjusted for sociodemographic and health factors including annual household income, education, age, sex, race, insulin use, BMI, and a number of other comorbid conditions. All analyses were conducted using Stata 8.0 (29) statistical software, with the SE adjusted for the clustering of patients by the VA Medical Center.

In our bivariate analyses, we found there was an association between chronic pain and other variables, such as depression and health status. Despite the potential collinearity among these variables, however, we believed for conceptual reasons that it was important to include all of these factors in the analysis. Moreover, the highest correlation among the individual variables was 0.3, and model diagnostics did not reveal any significant problems with including depression and health status, in addition to chronic pain, in the models.

The amount of missing data were generally 5% or less for most of the variables used in the statistical analysis. However, the amount missing was closer to 7% for three variables (chronic pain, number of comorbid conditions, and income). Consequently, we also analyzed the data using a multiple imputation procedure (30) to examine the potential influence of missing data on our study results. No substantial changes were observed for our key variable of interest, so we present the results of the nonimputed analyses only.

Approximately 60% of respondents (557 of 928) reported experiencing pain that was present most of the time for 6 months or more during the past year. The characteristics of patients with chronic pain versus those with no chronic pain are presented in Table 1. In general, patients reporting chronic pain compared with those without chronic pain were younger, included a higher proportion of women, were more likely to be using insulin, and had a higher BMI. A significantly higher percentage of those with chronic pain reported their health was fair or poor (55 vs. 33%; P < 0.001) and had a CES-D 10 score of 10 or greater (49 vs. 20%; P < 0.001), indicating the presence of significant depressive symptoms. On the other hand, individuals with chronic pain did not have a greater number of other chronic conditions than those without pain nor were they more likely to indicate that taking care of their diabetes was not a priority.

Among patients with chronic pain, almost one-third indicated that during the past 4 weeks their pain was severe or very severe, and 78% reported using pain medication on either a regular or occasional basis. The most commonly reported pain locations included the back (60%) and hip or knee (60%). In addition, patients indicated that, on average, pain had interfered with their daily activities on 18 of the last 28 days.

As shown in Table 2, the presence of chronic pain was significantly associated with poorer overall diabetes self-management (P = 0.002), even after adjusting for the presence of depressive symptoms, general health status, number of other comorbid conditions, and priority given to diabetes care. The logistic results suggest that chronic pain was most strongly associated with specific diabetes self-care activities, including greater difficulty with exercise (adjusted OR 3.0 [95% CI 2.1–4.1]) and following a recommended eating plan (1.6 [1.2–2.1]). The presence of chronic pain did not appear to be a significant barrier to taking diabetes medications, blood glucose monitoring, or checking feet for wounds or sores. In contrast, a positive depression screen was significantly associated with increased difficulty in taking diabetes medications, checking feet, and following a recommended eating plan.

Among patients with chronic pain, overall self-management was also significantly poorer for those who indicated their pain was severe or very severe compared with those who rated their pain as mild or moderate (P = 0.003) (Table 3). Specifically, patients with severe pain reported more difficulty with taking diabetes medications (adjusted OR =2.0 [95% CI 1.2–3.4]) and with exercise (2.5 [1.3–5.0]). For patients with chronic pain there was also a significant positive association between taking pain medication and reported diabetes self-management (P = 0.003). However, while those who reported taking pain medication had better diabetes self-management scores than patients not taking pain medication, their scores were still significantly lower than patients without chronic pain (79 [SD = 17] vs. 72 [15] vs. 67 [16], respectively; P < 0.01).

This study suggests that chronic pain is a prevalent condition among patients with diabetes. Furthermore, the presence of comorbid chronic pain and greater pain severity were both associated with poorer overall diabetes self-management and increased difficulty with certain self-care activities, such as exercising on a regular basis. These findings emphasize the importance of considering potential competing demands, such as chronic pain, when developing self-care regimens for patients with diabetes and other chronic health conditions that require high levels of self-management. Although innovative self-management programs for patients with a wide array of chronic conditions and comorbidities are being developed (31,32), a disease-specific focus continues to be the primary method for self-management education. The results from this study indicate that a more broadly focused approach that equips patients with the skills and confidence to manage chronic illnesses in general may be more effective in improving patients’ diabetes-related self-care.

Our findings also suggest that treatment of chronic pain (i.e., taking pain medication) may improve diabetes self-management, although, at least in this population, it does not completely eliminate the negative effect of chronic pain on self-management. However, taking pain medication is only a proxy measure. This project did not assess the types of pain medications being used, other pain control strategies (e.g., acupuncture), or the effectiveness of treatment for chronic pain. Thus, future studies are needed to explicitly address whether, and if so how, better pain control improves the performance of self-care behaviors and perhaps even more importantly diabetes clinical outcomes and patient quality of life.

We also did not assess the potential health consequences of comorbid chronic pain and poorer diabetes self-management. However, in a prior VA diabetes study, a higher score on the same self-management scale was associated with lower A1c levels, with a self-management score of 70 corresponding to a predicted hemoglobin A1c level of 7.7% and a self-management score of 30 corresponding with an A1c level of 8.3% (24). In the current study, as shown above, the self-management scores for patients with chronic pain were substantially lower than those for patients without chronic pain, which suggests the potential for poorer glycemic control among this subgroup of patients. Nonetheless, the effect of chronic pain on the achievement of good glycemic, lipid, and blood pressure control requires additional research.

This study further highlights the often observed link between pain and depression (8) and between chronic pain and poor self-rated health (26). Moreover, it suggests there is an independent association between chronic pain and self-management even after these relationships are taken into account. Similar to prior research (25), we found that a positive depression screen affects patients’ performance of certain activities, such as taking diabetes medications and following an eating plan. On the other hand, our results did not show a significant association between depressive symptoms and exercise, as found in previous work (25), but instead revealed that both the presence of chronic pain and pain severity had a significant effect on patients’ reported difficulty with exercising.

There are some study limitations that need to be discussed. First, while we found that the majority of patients with chronic pain reported pain involving back and knees, we did not assess the types of pain conditions affecting our study population. However, the presence of peripheral neuropathy, which may produce pain, was abstracted as part of the baseline medical record review. According to these data, about 19% of the study population had a diagnosis of peripheral neuropathy (21% in the chronic pain group and 16% among those who reported no chronic pain), and thus peripheral neuropathy does not appear to be the only pain condition influencing our study results. Second, the response rate to the initial survey limits the generalizability of these study results. Therefore, it will be important to conduct additional studies with more representative samples to determine whether these results can be replicated both inside and outside the VA system. Finally, we excluded from our analyses 65 individuals who did not respond to the chronic pain question. To examine whether this decision affected our results, we ran our main analyses first including these individuals in the no chronic pain group and then including them in the chronic pain group. Neither of these approaches produced a substantial change in the results.

In conclusion, the rapidly growing number of individuals with diabetes and the potential for poor outcomes without proper management only serve to heighten the importance of providing optimal diabetes care and improving patient self-management. One possibly significant but generally overlooked aspect of diabetes care is how other chronic conditions, such as chronic pain, might complicate and/or serve as a competing demand in a patient’s ability to engage in self-care activities. This study suggests that comorbid chronic pain may be a major limiting factor in the performance of certain self-care behaviors and thereby reinforces the need to proactively address such potential competing demands as we seek to support and improve patients’ diabetes self-management.

This study was supported through grant funding from the Department of Veterans Affairs, Health Services Research and Development Service, SDR 01-019 and DIB 98-001, and used core support from the Michigan Diabetes Research and Training Center funded by NIH5060 DK20572.

A special thank you goes to Emily Lipp for helping with data management.