OUP user menu

Cognitive–Behavioral Therapy for Chronic Low Back Pain in Older Persons: A Preliminary Study

M. Carrington Reid PhD, MD, John Otis PhD, Lisa C. Barry MPH, Robert D. Kerns PhD
DOI: http://dx.doi.org/10.1046/j.1526-4637.2003.03030.x 223-230 First published online: 1 September 2003


Objective. To determine the feasibility and potential efficacy of providing cognitive–behavioral therapy (CBT) to older persons with chronic low back pain (CLBP).

Methods. This was an uncontrolled pilot study conducted at a senior housing center (SHC) in New Haven, Connecticut. Fourteen SHC residents aged 65 years and older who were cognitively intact (Mini Mental State Examination score ≥24) and had CLBP were recruited for the study. CBT was administered in 10 weekly individual sessions. Participants were phoned 5 days on average after each session (range: 3–7 days) to determine their comprehension and perceived usefulness of the CBT materials and adherence with the assigned homework exercises. Using standardized measures, we determined participants' levels of pain intensity, pain-related disability, and physical and social activity at baseline, and at 2 and 24 weeks posttreatment.

Results. Participants had a mean age of 77.4 (±7.9 SD) years and were mostly female (86%). Thirteen (93%) participants completed all 10 sessions. Comprehension of CBT, defined as self-reported understanding of the materials presented each week, exceeded 97%. The perceived usefulness of each treatment session was assessed on a 0–10 scale, and the mean ratings for the sessions ranged from 7.5–9.4. The mean number of days that participants practiced the homework exercises each week varied from 1.8 to 4.0. Significant reductions (P < 0.01) in participants' pain intensity and pain-related disability scores were found at the 2-week posttreatment (vs pretreatment) assessment. These treatment effects waned over time, but did not return to pretreatment levels at 24 weeks. Participants' physical and social activity levels did not change.

Conclusions. CBT is a feasible treatment for cognitively intact, older persons with CLBP, and may be efficacious as well.

  • Chronic Low Back Pain
  • Cognitive–Behavioral Therapy
  • Older Persons


Chronic low back pain (CLBP) is a common health problem among older persons [1] that is often associated with substantial disability [2–4]. Despite the prevalence of and disability associated with CLBP in older populations, effective treatment strategies remain inadequately defined. Although analgesic medications are commonly used to treat CLBP [5], pharmacologic options are often limited (particularly among older persons with substantial comorbidities) given the side-effect profiles of and risks associated with many analgesic agents, including nonsteroidal anti-inflammatory [6,7] and opioid medications [8,9]. Developing effective nonpharmacologic treatments could possibly provide substantial benefit to many older persons with CLBP.

Cognitive–behavioral therapy (CBT) is a potentially effective treatment for reducing pain and improving function in older persons with CLBP [10]. CBT seeks to enhance a person's ability to cope effectively with pain and its consequences using a variety of cognitive and behavioral methods [11]. The therapy is typically limited to 10 weekly treatment sessions and is administered either individually or in a group format. Prior studies of nonelderly persons with chronic pain have shown that CBT can reduce levels of pain and pain-related disability, as well as enhance a person's ability to self-manage pain [12]. To our knowledge, the feasibility and efficacy of CBT have not been previously examined in a study of older persons with CLBP. Demonstrating the feasibility of providing CBT to older persons with CLBP is an important first step in evaluating its potential utility. Older persons with chronic pain may be reluctant to try nonpharmacologic treatments [13]. Attendance at the weekly therapy sessions could be compromised by functional impairments, which are common among older persons with CLBP [2–4]. Furthermore, CBT employs weekly homework assignments as a means of reinforcing key concepts; however, the extent to which older persons affected by CLBP may comply with this aspect of the therapy is unknown. Finally, the presence of chronic pain has been associated with attention and memory deficits [14,15], which could, if present, impair older persons' comprehension and/or retention of CBT materials. These effects may be compounded among older persons with age-related memory declines.

Accordingly, our primary objective was to determine the feasibility of providing CBT to older persons with CLBP. We provided CBT in individual, as opposed to group, sessions to maximize the tailoring of the therapy, given the potential for individual differences between participants. We adapted a previously developed CBT treatment protocol [16] for use in the current study. As a secondary objective, we examined the potential efficacy of CBT.



Participants were residents of a senior housing center (SHC) in New Haven, Connecticut, recruited by local advertisements at the SHC and by clinician referral. Primary care clinicians who staffed a clinic located at the SHC (which provides longitudinal care for approximately 800 of the 1,200 residents) were asked to provide the names of patients they felt were appropriate to participate. A total of 33 potential participants were identified using these recruitment strategies.

A two-step screening process was used to ascertain each person's eligibility status. First, we telephoned the 33 potential participants to determine if they experienced low back pain due to a noncancer cause on most days of every month for at least 3 months in the past year, were aged 65 years or older, and spoke English. We successfully contacted 31 (94%) people by phone: Five were not interested in participating, four were ineligible, and the remaining 22 met the above eligibility criteria. All 22 potentially eligible persons were asked to provide oral consent to undergo a face-to-face evaluation at the SHC clinic. Individuals were informed that they would receive $25 as compensation for the time spent (approximately 45 minutes) undergoing this assessment.

Of the 22 persons who were deemed potentially eligible to participate, 16 provided oral consent and underwent an evaluation that included the battery of measures described below (Step 2). Additional inclusion criteria employed at the time of this face-to-face assessment included a Mini Mental State Examination [17] score of 24 or greater and the presence of back pain on most days of every month during the 3-month period prior to the assessment. Because CLBP is often a remitting and recurring disorder, we required that participants met the latter criterion to ensure that all of them had back pain at the time of the study. All 16 people who underwent the face-to-face assessment were eligible to participate; 14 provided written consent and enrolled in the study. The study took place between February 2002 and November 2002 and was approved by the Human Investigation Committee at Yale University. Participants were compensated ($20 per treatment session) for their participation.

Baseline Assessment

Information regarding participants' demographic characteristics (age, gender, race, educational level, and marital status) was obtained during the baseline assessment by a trained interviewer. We assessed for 12 self-reported, physician-diagnosed chronic conditions: Hypertension, myocardial infarction, congestive heart failure, arrhythmia, diabetes, stroke, cancer (other than minor skin cancers), cirrhosis or liver disease, chronic lung disease, arthritis, osteoporosis, and Parkinson's disease. Depressive symptomatology was assessed using the 11-item Center for Epidemiologic Studies Depression (CES-D) scale [18]. We determined the duration and self-reported cause(s) of low back pain, as well as the number of pain sites (in addition to low back) for each participant. Using an open-ended question, we asked participants to name the methods they used to manage their CLBP at the time of enrollment.


Because our primary objective was to determine the feasibility of providing CBT to older persons with CLBP, all participants received the intervention, which consisted of 10 weekly sessions of individual therapy (lasting approximately 50 minutes each) administered by a psychologist with expertise in CBT. All treatment sessions took place in office space provided by the SHC clinic. Treatment was based on a previously developed pain management treatment protocol [16] that emphasized identifying and modifying maladaptive thoughts and behaviors related to the experience of chronic pain. While some modifications were made to the treatment manual to tailor the information and homework exercises to an older group (e.g., use of age-appropriate examples and larger font on handouts), the core components of CBT for pain were provided.

Specific topics covered during the treatment sessions included identification of participants' beliefs about pain and pain treatment, as well as reconceptualization of the pain experience as subject to personal control (Session 1). Participants were educated about the various theories of pain and were taught relaxation techniques including diaphragmatic breathing, progressive muscle relaxation, and visual imagery (Sessions 2–4). Participants were taught a time-based pacing technique in which individuals take breaks based on time or amount of work accomplished rather than pain level, and were educated regarding the importance of scheduling pleasant activities (Sessions 5–6). Cognitive restructuring was taught to help participants learn to identify maladaptive thoughts and beliefs related to pain and substitute more adaptive ones (Session 7). Additional training in techniques related to anger management and sleep hygiene (Session 8), as well as relapse prevention (Sessions 9–10), was also provided. As a method of reinforcing the presented materials, participants were given homework assignments at the end of each of Sessions 1–9. Typically, they were asked to practice the various techniques presented during the previous session. The therapist also collaborated with each participant to generate specific intersession goals (e.g., daily walks for 20 minutes) after each of Sessions 1–9. The format of Sessions 2–9 included: 1) A review of the homework assignment and intersession goal accomplishment; 2) Problem-solving discussions regarding nonadherence with the homework assignments and/or difficulties faced accomplishing the intersession goals; 3) A brief review of materials covered in the previous session; 4) Presentation and practice of new skills; and 5) Collaborative establishment of new homework/goals for the next week. The cognitive and behavioral techniques presented in Sessions 1–9 were reviewed in Session 10.

Feasibility Measures

The research assistant telephoned participants 5 days, on average, after the end of the sessions (range: 3–7 days) and administered the feasibility measures. To ascertain comprehension of the CBT materials presented during Sessions 1–9, participants were asked, “Did you understand the material presented at the previous session?” (yes/no/don't know). Because it was a review session, we did not assess participants' comprehension of the materials presented at Session 10. We measured comprehension to determine whether the CBT materials should be further modified prior to subsequent studies. Participants' perceived usefulness of the individual treatment sessions was assessed by asking, “How useful was the material presented in last week's session?” Responses were recorded using a 0 (not at all useful) to 10 (extremely useful) scale.

We evaluated participants' adherence with the homework exercises by asking them if they completed the homework assignments after Sessions 1 and 9 (yes/no) and if they practiced the specific CBT techniques following Sessions 2–8 (yes/no). Those answering yes to the latter question were asked to report the number of days they practiced the technique since the last treatment session. (At the end of Session 1, participants were asked to complete a worksheet identifying things that made their pain increase or decrease; after Session 9, they were asked to practice any CBT technique of their choice.)

Efficacy Measures

The following measures were administered at baseline and during a face-to-face follow-up evaluation at 2 weeks posttreatment and a telephone interview at 24 weeks posttreatment. Average pain intensity over the previous 4-week period was measured using a 0–10 numeric-rating scale [19]. Pain-related disability measures included: 1) The 24-item Roland-Morris Disability Questionnaire [20], which assesses the extent to which back pain interferes with a person's ability to perform (or limits the time spent doing) basic tasks such as walking, transferring, and dressing (score range: 0–24); 2) Days of restricted activity due to back pain in the past month [21], where response categories included 0, 1–7, 8–14, and 15+ days; and 3) Pain-related interference with basic activities of daily living (BADL) and instrumental activities of daily living (IADL). Participants were asked to rate the degree to which pain interfered with their ability to do seven BADL and seven IADL [22] during the past month, using a three-item response scale [23] that ranged from not at all (score 0) to a lot (score 2). Participants' responses were summed to yield total BADL (range: 0–14) and IADL (range: 0–14) pain-related interference scores, with higher scores indicating greater disability. Test-retest reliability of the BADL and IADL pain-related interference measures was substantial in our sample, as evidenced by intraclass correlation coefficients of 0.82 and 0.78, respectively.

Participants' levels of physical activity during the previous month were assessed using a previously developed nine-item physical activity questionnaire [24]. We administered five of the nine items that were felt to be appropriate for our sample, including light housework, heavy housework, vigorous sports such as swimming, less vigorous sports such as bowling, and other leisure time or sporting activities. Responses were coded using a four-item response scale: Not at all (score 0), less than once a month (score 1), 1–4 times a month (score 2), and more than four times a month (score 3). A total score (range: 0–15) was calculated for each participant, with higher scores reflecting greater physical activity. Participants' levels of social activity during the previous month were ascertained using a protocol [25] adapted from the Epidemiologic Studies of the Elderly (EPESE) interview [26]. The frequency with which participants engaged in 10 types of social activities (e.g., visiting friends, attending religious services, participating in group activities, etc.) was determined using the four-item response scale described above. A total score (range: 0–30) was calculated for each participant, with higher scores reflecting greater social activity.

To assess for change in participants' levels of depressive symptoms, we readministered the 11-item CES-D questionnaire at both follow-up assessments.


The univariate distributions of all study variables were initially inspected for coding errors and maldistributions. We assessed for differences between participants' pretreatment and post-treatment scores on the efficacy measures using Wilcoxon Signed Rank tests. Two-tailed P values of less than 0.05 were considered statistically significant.


Participants had a mean age of 77. 4 (±7.9 SD) years, and were mostly female (86%) and Caucasian (79%). The mean educational level of the sample was 11.4 (±3.0 SD) years, and few participants were currently married (21%). The median duration of back pain in years was 12.0 (range: 0.3–43.5) and the mean number of pain sites including low back was 2.6 (±1.4 SD). Participants used an average of 2.1 (±0.8 SD) methods to manage their CLBP at the time of enrollment, including analgesic medications, activity restriction, and stretching/exercise by 79%, 29%, and 29% of participants, respectively.

Thirteen (93%) participants completed all 10 treatment sessions. One participant dropped out after Session 2 because she did not want to do the homework exercises. A total of 117 interviews (13 participants × 9 interviews + 1 participant × 2 interviews) were conducted to determine participants' comprehension of the CBT materials. The total numbers of yes, no, and don't know responses were 114, 0, and 3 (one after Session 2 and two after Session 9), respectively, indicating a high degree (114/117=97%) of self-reported comprehension of the CBT materials.

Table 1 shows the mean, standard deviation, and range of perceived usefulness ratings for the 10 treatment sessions. The mean ratings on the 0–10 usefulness scale for the sample ranged from a low of 7.5 (±2.4 SD) for Session 7 to a high of 9.4 (±0.9 SD) for Session 10. The mean number of days that participants practiced the homework exercises varied from 1.8 (±1.3 SD) for Session 6 to 4.0 (±2.3 SD) for Session 8 (Table 1).

View this table:
Table 1

Perceived usefulness of the individual treatment sessions and number of days homework exercises were practiced

Perceived usefulness ratings*Number of days homework
exercises were practiced
Mean (SD)RangeMean (SD)Range
 1 Theories of pain8.1 (1.7)5–10100%
 2 Diaphragmatic breathing9.0 (1.5)5–102.7 (2.0)0–7
 3 Progressive muscle relaxation8.3 (1.6)5–103.1 (1.3)2–6
 4 Visual imagery8.8 (2.4)2–102.3 (1.7)0–5
 5 Timed-based pacing8.5 (1.6)5–102.8 (2.1)0–7
 6 Pleasant-activity scheduling9.5 (0.9)7–101.8 (1.3)0–4
 7 Cognitive restructuring7.5 (2.4)2–102.0 (2.1)0–7
 8 Anger management and sleep hygiene9.1 (0.9)8–104.0 (2.3)0–7
 9 Relapse prevention9.2 (0.9)8–10100%
10 Review session9.4 (0.9)7–10NA§
  • * Ratings could vary from 0 (not at all useful) to 10 (extremely useful).

  • The research assistant interviewed participants 5 days, on average, after each session (range 3–7 days).

  • Adherence was calculated as the percentage of participants who reported doing the exercise (yes/no). At the end of Session 1, participants were asked to complete a worksheet identifying things that made their pain increase or decrease, and following Session 9, they were asked to practice any CBT technique of their choice.

  • § NA=not assessed.

Significant reductions in participants' pain intensity and pain-related disability scores were found at 2 weeks posttreatment (Table 2). These treatment effects waned over time but did not return to pretreatment levels at 24 weeks. We did not observe a change in participants' physical and social activity levels.

View this table:
Table 2

Pretreatment, and 2-week and 24-week posttreatment scores

Pretreatment (N=14)2 weeks Posttreatment (N=13)P value*2 weeks Posttreatment (N=13)P value*
Average intensity (0–10), mean (SD) 7.4 (2.0) 3.9 (2.5)0.009 5.5 (2.9)0.129
Pain-related disability
Roland-Morris score (0–24), mean (SD)14.4 (4.4) 8.5 (6.4)0.00210.4 (7.4)0.052
Interference with BADL (0–14), mean (SD) 6.0 (2.6) 1.9 (2.4)0.001 3.8 (4.2)0.123
Interference with IADL (0–14), mean (SD) 5.1 (3.0) 2.5 (2.6)0.005 3.4 (3.1)0.135
Days of restricted activity due to back pain, N (%)
0 1 (7) 5 (38)0.008 4 (31)0.039
1–7 4 (29) 7 (54) 5 (39)
8–14 3 (21) 1 (8) 2 (15)
15+ 6 (43) 0 (0) 2 (15)
Physical and social activity
Physical activity score (0–15), mean (SD) 8.6 (1.8) 7.7 (3.1)0.215 7.9 (3.3)0.569
Social activity score (0–30), mean (SD)16.6 (5.7)12.4 (6.5)0.06014.5 (5.3)0.242
  • * The P value refers to the difference between the baseline and follow-up scores as determined by Wilcoxon signed rank tests.

  • Higher scores reflect greater disability on the Roland-Morris and pain-related interference with BADL and IADL measures.

  • Higher scores reflect greater levels of physical and social activity.

The mean CES-D score for the sample was 17.9 (±11.7 SD) at baseline and decreased to 10.5 (±9.4 SD) and 13.8 (±9.9 SD) at 2 and 24 weeks posttreatment (P=0.005 and 0.049), respectively.


Our pilot study demonstrates the feasibility of providing CBT to cognitively intact, older persons with CLBP. To our knowledge, this is the first study to examine the use of CBT in older persons with CLBP and represents an important first step in evaluating the potential utility of this particular therapy.

We employed a treatment protocol that had been originally developed and standardized for use in younger, treatment-seeking persons with chronic pain. We modified the protocol slightly (e.g., intersession goals were tailored for an older group) and assessed participants' comprehensions of the materials to determine whether additional modifications would be necessary prior to subsequent studies. There was a high rate of self-reported comprehension of the CBT materials among participants, which suggests that standard CBT pain protocols can be administered (with few modifications) to cognitively intact, older persons with CLBP. Furthermore, participants rated the treatment sessions as highly useful and reported doing the weekly exercises, such as practicing visual imagery or muscle relaxation techniques. These findings suggest that older persons with CLBP are willing to try CBT methods and may find them useful.

Although our results were not obtained in the context of a randomized clinical trial, our pilot data suggest that CBT may be an efficacious therapy for CLBP in older persons. We found significant reductions in participants' pain intensity and pain-related disability scores at 2 weeks posttreatment. Participants' pain intensity scores decreased, on average, by three points on the 0–10 numeric-rating scale. Prior research suggests that a reduction of two or more points on this 11-item scale represents a clinically meaningful reduction in pain intensity level [27]. A reduction of 2–3 points on the 24-item Roland-Morris scale has been previously recognized as the minimum degree of change that is considered clinically important [21]. We observed an average decrease of approximately six points in participants' Roland-Morris scores at the 2-week posttreatment evaluation. The observed reductions in days of restricted activity due to back pain and pain-related interference with BADL and IADL functions were also substantial. Of course, we cannot be certain that the observed treatment effects were due strictly to the intervention, given the absence of a control group. It is possible that the reductions in pain intensity and pain-related disability were simply due to a regression to the mean. Our results are consistent, however, with the findings of the few studies that assessed the efficacy of CBT among community-living older persons with osteoarthritis of the knee [28] and nursing home residents with diverse chronic pain conditions [29].

We did not observe a significant change in participants' physical and social activity levels at follow-up. It is possible that our measures were not sufficiently sensitive to detect a change in these outcomes during the study period or that improvements in physical and social activity take longer than 24 weeks to occur. Alternatively, reducing levels of pain and pain-related disability in basic functional tasks may have little impact on older person's physical and social activity levels if other factors (e.g., deconditioning, social isolation, fear of pain exacerbation outside the home, etc.) are also present. This finding highlights the need for future investigations that improve our understanding of the barriers to improving physical and social function in older persons with CLBP.

We found significant reductions in participants' depressive symptom scores at 2 and at 24 weeks posttreatment. The presence of depressive symptoms has been shown previously to be a predictor of disabling musculoskeletal pain among community-living older persons [30]. Future research is needed to explore the potential mediating versus moderating role of depressive symptoms in this setting.

Our study has several important limitations that warrant comment. Because we used self-reported measures to ascertain the feasibility of providing CBT, we cannot exclude the possibility that participants overreported their comprehension and perceived usefulness of the treatment sessions, as well as their adherence to homework exercises. With the exception of our pain-related interference measure, we did not assess the reliability or validity of our outcome measures. In addition, we assessed for maintenance of treatment effects over a relatively limited period of time (i.e., 6 months posttreatment). Furthermore, we did not determine participants' compliance with the various CBT techniques at the follow-up evaluations. Finally, the absence of transportation barriers and the financial compensation that individuals received for participation may have affected our ability to recruit (and retain) participants.

In conclusion, our pilot study demonstrates that CBT is a feasible and potentially efficacious treatment for cognitively intact, older persons with CLBP. These data provide strong support for a randomized controlled trial to determine whether CBT can reduce levels of pain and pain-related disability, improve physical and social functioning, as well as decrease reliance on pharmacologic treatments among older persons affected by CLBP.


  • This study was funded through a grant from the Claude D. Pepper Older American Independence Center (PGOAG10469).


View Abstract