JAGS 50:2029–2034, 2002© 2002 by the American Geriatrics Society 0002-8614/02/$15.00

Is Satisfaction With Pain Management a Valid and Reliable Quality Indicator for Use in Nursing Homes?

David J. Casarett, MD, MA,*

†‡§

Karen B. Hirschman, PhD, MSW,

†

�

Elizabeth R. Miller, BS,

‡

and John T. Farrar, MD

¶

OBJECTIVES:

To determine whether satisfaction withpain management can be measured reliably in nursinghomes and to gather preliminary data about the validity ofsatisfaction assessments in this population.

DESIGN:

Cross-sectional interview study.

SETTING:

Two urban nursing homes.

PARTICIPANTS:

Sixty-six nursing home residents withpain.

MEASUREMENTS:

Overall satisfaction with pain manage-ment, satisfaction with pain medication, experiences relatedto pain management, cognitive function, depressive symp-toms, and retest reliability of overall satisfaction rating.

RESULTS:

Most residents (60/66; 91%) could rate their over-all satisfaction with pain management. Overall satisfactionwas weakly correlated with pain severity at the time of the in-

terview (Spearman correlation coefficient

�

–0.28;

P

�

.033)and over the past week (–0.27;

P

�

.038). Overall satisfac-tion was also negatively associated with the GeriatricDepression Scale score (–0.50;

P

�

.001). Satisfaction withpain medication was associated with several ratings of themedication’s beneficial effects, including improved activ-ity, sleep, and speed of relief, but not with the frequency withwhich it caused side effects. Ratings of overall satisfactionshowed good reliability overall (kappa

�

0.62;

P

�

.001)and for those with Mini-Mental State Examination scoresgreater than 21 (kappa

�

0.70;

P

�

.001) and 21 or less(kappa

�

0.54;

P

�

.004).

CONCLUSION:

These results suggest that satisfactionwith pain management can be measured reliably when res-idents are able to report their pain, but further research is

needed before satisfaction with pain management can beincorporated into routine assessments in nursing homes.

JAm Geriatr Soc 50:2029–2034, 2002.

Key words: pain management; satisfaction; nursing homes

C

hronic pain is common in nursing home residents,and, although prevalence figures vary,

1–3

some esti-mates are as high as 80%.

4

Moreover, chronic pain is re-sponsible for significant decreases in quality of life and hasa broad effect on independence, relationships, and activ-ity.

2,5,6

It is not surprising, therefore, that nursing homesare increasingly examining their ability to manage pain ad-equately.

7

Satisfaction with pain management is one method ofevaluating pain management that has been widely employedin non-nursing home settings.

8–10

The American GeriatricsSociety Task Force on pain management has recommendedthat institutions make satisfaction with pain management apriority in quality-improvement efforts,

11

but several ques-tions must be answered before satisfaction with pain man-agement can be used as a quality indicator in nursing homes.

First, it is important to determine whether it is possibleto reliably measure nursing home residents’ satisfaction withtheir pain management. Cognitive impairment is commonin nursing home residents

12

and may influence their per-ceptions of pain and their reporting of pain severity.

2,4

Therefore, satisfaction measures may not be sufficientlyreliable in this population.

Second, it is equally important to ensure that mea-sures of satisfaction are valid. At a minimum, satisfactionshould exhibit a strong inverse correlation with pain sever-ity. This criterion of validity is particularly challenging,because, although some studies in other settings havefound this relationship,

13,14

others have not.

8,9

These latterfindings suggest that there may be a variety of other fac-tors, like speed of relief or the presence of side effects, thatalso influence satisfaction. Therefore, residents’ satisfac-tion with treatment should also be positively correlatedwith their perceptions of important outcomes of treat-ment, such as decreased pain or improved activity.

From the *Philadelphia VA Medical Center, Philadelphia, Pennsylvania; and

†

Division of Geriatric Medicine,

‡

Center for Bioethics,

§

Leonard Davis Institute

for Health Economics,

�

School of Social Work, and

¶

Center for Clinical Epidemi-ology and Biostatistics, University of Pennsylvania, Philadelphia, Pennsylvania.

David Casarett is the recipient of a Research Career Development Award in Health Services Research from the Department of Veterans Affairs and by a Paul Beeson Physician Scholars Award. This study was supported by the Commonwealth Fund.

Address correspondence to David J. Casarett, MD, MA, 3615 Chestnut Street, Philadelphia PA 19104. E-mail: casarett@mail.med.upenn.edu

2030

CASARETT ET AL.

DECEMBER 2002–VOL. 50, NO. 12 JAGS

This study was conducted to determine whether satis-faction with pain management can be measured reliably innursing homes and to gather preliminary data about thevalidity of satisfaction assessments in this population.

METHODS

Subjects

Nursing home residents were recruited from a 240-bedVeterans Affairs (VA) long-term care facility and a 120-bed facility affiliated with the University of Pennsylvania.These facilities were selected in order to recruit a diversesample of residents. The principal investigator (DC)trained the interviewer, who approached each residentonce. The interviewer screened potential subjects by ask-ing whether they had “any pain the last 7 days.” Residentswere eligible for this study if they reported to the inter-viewer that they had pain in the week before the interviewand if they could respond appropriately to interview ques-tions. Residents who refused were not approached again.

We estimated that both facilities combined (n

�

360)would provide at least 50 residents who met inclusion crite-ria. A sample size of 50 provides adequate power (0.90) todetect a correlation of 0.40 (Pearson product-moment coeffi-cient) between satisfaction and pain severity (alpha

�

0.05).

15

Design Procedures

At the larger facility, nurse practitioners helped to excludethose residents who were too cognitively impaired to re-spond to the interview questions. At the smaller facility,the interviewer approached all residents. After oral in-formed consent was obtained, residents described the loca-tion of their pain and its severity over the past week and atthe time of the interview. A verbal descriptor scale from 0to 7 was used,

16

because previous work has shown thatnursing home residents may be better able to completescales with verbal descriptors and fewer response options.

2

The satisfaction portion of the interview had threequestions. First, residents rated their overall satisfactionwith the way that their pain was being treated (overall satis-faction), from “very dissatisfied” to “very satisfied,” using a6-point scale. This item is reproduced in Appendix 1.Those who reported that they were receiving pharmaco-logical or nonpharmacological therapy were asked to ratetheir satisfaction with each, using the same scale.

For each type of treatment, residents rated the frequencyof positive and negative experiences they attributed to it,from “not at all” to “all of the time,” using a 6-point scale.For instance, residents were asked how often their painmedication helped them to be more active and how oftenit made them drowsy. There were nine questions for medi-cation therapy and for nonpharmacological therapy.These ratings were modeled after strategies employed inother surveys of satisfaction, which supplement overallratings of satisfaction with measures of specific experi-ences or events.

17,18

All residents were also asked to ratethe frequency with which nursing home staff asked abouttheir pain and the frequency with which staff appear to betrying to treat it.

The interviewer used cards that displayed the responseoptions in large print to reinforce verbal scripted ques-tions, unless subjects said that they were unable to read

them. The interviewer also administered the Folstein Mini-Mental State Examination (MMSE)

19

and the GeriatricDepression Scale (GDS) 15-item short form.

20–22

At theconclusion of the interview, residents’ charts were re-viewed to obtain demographic data and informationabout medication use and nonpharmacological therapy forpain. The Institutional Review Boards for both nursinghomes approved this study.

Statistical Analysis

The retest reliability of the overall satisfaction measurewas evaluated by reassessing satisfaction, current pain,and pain over the past week. Other questions were not re-peated. Repeat interviews were conducted as close as pos-sible to 3 days after the initial interview.

The associations between overall satisfaction and medi-cation satisfaction and their respective experiential ratingswere assessed using the Spearman correlation coefficient. Theassociation between satisfaction and dichotomous variableswas evaluated using the Wilcoxon rank sum test, which ismathematically equivalent to the Mann-Whitney

U

test. Toadjust for possible confounding, selected variables thatwere associated with overall satisfaction in bivariate anal-ysis (pain severity and GDS score) were entered into an or-dinal logistic regression equation, in which satisfactionwas the outcome variable. The test-retest reliability of theoverall satisfaction measure was examined using a kappastatistic with quadratic weighting, which is equivalent tothe intraclass correlation coefficient.

23

Stata software(Stata for Windows, Version 5.0; College Station, TX) wasused for all data analysis.

RESULTS

Three hundred forty-seven residents were identified on aninitial bed review; of these, 230 had sufficient cognitive ca-pacity to participate in an interview. Eighty-two of theseresidents had pain in the past week (82/230; 36%); of these,66 agreed to participate (66/82; 80%). Their characteristicsare described in Table 1. There was no difference in the re-cruitment rate in the two nursing homes. Interviews gener-ally took 30 to 40 minutes to complete.

Most residents could rate their overall satisfactionwith their pain management (60/66; 91%). Those who didnot appeared unable to choose a response option. Mostresidents who could answer this question reported thatthey were satisfied (median

�

5 “satisfied”; range 1–6)(Table 2). However, residents believed that nursing homestaff did not always try to manage their pain (median

�

3“about half the time”) and that staff infrequently askedabout their pain (median

�

1 “rarely”) (Table 2). There wasno difference between the MMSE scores of those who couldrate their overall satisfaction and those who could not.

Overall satisfaction showed a weak but significantinverse correlation with pain severity at the time of the inter-view (Spearman correlation coefficient

�

�

0.28;

P

�

.033)and over the past week (Spearman correlation coefficient

��

0.27;

P

�

.038). (Table 3) Among residents with mild pain(1–4) over the past week and at the time of the interview,there was no correlation between pain and satisfaction (pastweek: Spearman rho

�

0.36,

P

�

.26; at interview: rho

��

0.09,

P

�

.58). However, among residents with severe

JAGS DECEMBER 2002–VOL. 50, NO. 12

SATISFACTION WITH PAIN MANAGEMENT

2031

pain (5–7) over the past week and at the time of the inter-view, pain was correlated with satisfaction (past week: rho

��

0.36,

P

�

.029; at interview: rho

�

�

0.73,

P

�

.001).Residents’ overall satisfaction was not related to their rat-

ings of how frequently nursing home staff asked about theirpain (Spearman correlation coefficient

�

0.25;

P

�

.051) butwas strongly correlated with their ratings of how rigor-ously staff tried to treat their pain (Spearman correlationcoefficient

�

0.51;

P

�

.001). Overall satisfaction wasalso strongly correlated with satisfaction with medication(Spearman correlation coefficient

�

0.73;

P

�

.001) andwas negatively associated with GDS score (Spearman cor-relation coefficient

�

�

0.50;

P

�

.001). This relationshipremained after adjusting for pain severity at the time of theinterview in an ordinal logistic regression equation (GDScoefficient

�

�

0.32,

P

�

.001; pain coefficient

�

�

0.99;

P

�

.044). Overall satisfaction was not associated withage, sex, ethnicity, MMSE score, or education.

Sixty-two residents were taking pain medication, butonly 56 were aware that they were (90%). Of these, allcould rate their satisfaction with their medication. Residentswere only moderately satisfied with their pain medication(median

�

4 “slightly satisfied”; range 1–6). Satisfactionwith pain medication was associated with several ratings ofthe medication’s effects, including improved activity, sleep,concentration, and speed of relief (Table 3). Because manyresidents had difficulty with ambulation for reasons unre-lated to pain (e.g., amputation) only 38 (68%) residentswere able to describe the effect of their medication on theirability to walk.

Residents’ satisfaction with their pain medication wasnot associated with the frequency with which they per-ceived that it caused side effects such as drowsiness andconstipation (Table 3). Nor was satisfaction associatedwith opioid use (4.1 vs 3.8; rank sum test

P

�

.49). Amongresidents receiving opioids, there was no difference in satisfac-tion between residents receiving scheduled (mean

�

4.5) andas needed only (mean

�

3.7) preparations (rank sum test

P

� .29). Satisfaction with pain medication was correlatedwith GDS score (Spearman rho � �0.38; P � .004) but notwith age, sex, education, MMSE score, or ethnicity.

Nine residents (14%) reported that they were receiv-ing nonpharmacological treatments for pain. Residents’assessments were often vague and were difficult to verifyby chart review. Most of those residents who reported re-ceiving a nonpharmacological therapy for pain were verysatisfied with its effectiveness (mean 5.3; range 4–6). Be-cause very few residents were aware that they were receiv-ing nonpharmacological treatment for pain, correlates ofsatisfaction with these treatments were not examined.

Follow-up interviews could be conducted with 57 res-idents, with a median interval of 5 days. Repeat ratings ofoverall satisfaction showed moderate agreement (kappa �0.62; P � .001) indicating good retest reliability. Therewas no significant change in pain severity at the time ofthe interview (initial mean � 2.4, follow-up mean � 2.0;sign rank test P � .27) or in the past week (initial and fol-low-up mean � 2.7; sign rank test P � .77). The effect of cog-nitive impairment on reliability was evaluated by splitting thesample at the median MMSE score (�21 and �21). Kappasfor these groups were 0.70 (P � .001) and 0.54 (P � .004),respectively.

Table 1. Resident Characteristics (N � 66)

Characteristic Value

Age, mean � SD (range) 72.7 � 12.8 (40–97)Sex, n (%)

Female 18 (27)Male 48 (73)

Marital status, n (%)Single 18 (27)Married 18 (27)Divorced/Separated 13 (20)Widowed 17 (26)

Ethnicity, n (%)African American 33 (50)White 33 (50)

Education, years, n (%) �12 30 (45)12 19 (29)13–15 12 (18)16 3 (5)�16 2 (3)

Pain location, n (%)*Back 19 (29)Hip/leg 15 (23)Feet 9 (14)Upper extremity 9 (14)Knee 8 (12)Diffuse 7 (11)Head/neck 5 (8)Abdomen 4 (6)Chest 3 (5)

Documented musculoskeletaletiology for pain (e.g., degenerative joint disease), n (%) 35 (53)

Pain medication, n (%)†

Nonopioid‡ 49 (79)Opioid 17 (27)Immediate release 9 (15)Sustained release 8 (13)

Documented diagnosis of majordepression, n (%) 15 (23)

Taking antidepressants, n (%) 25 (38)Number of pain medications,

mean � SD (range) 2.0 � 1.0 (0–4)Pain severity, mean � SD (range)At time of interview 2.4 � 2.5 (0–7)Over past week 4.7 � 1.4 (1–7)Mini-Mental State Examination,

mean � SD (range) 20.0 � 6.1 (5–30)Geriatric Depression Scale,

mean � SD (range) 5.6 � 4.4 (0–14)Geriatric Depression Scale

score �4, n (%) 38 (58)

*Percentages do not add to 100 because some residents described pain in multiplesites.†Percentages do not add to 100 because some residents received more than onemedication/nonpharmacological therapy.‡Nonsteroidal antiinflammatory agents, acetaminophen.SD � standard deviation.

2032 CASARETT ET AL. DECEMBER 2002–VOL. 50, NO. 12 JAGS

DISCUSSIONAs pain management in nursing homes receives increasingattention, it will be important to better understand howpain and outcomes of treatment can be measured in thispopulation. In particular, it will be important to determine

whether satisfaction, one of the most widely used outcomemeasures, is a valid and reliable measure in nursing homeresidents. The results of this study have several implica-tions for the feasibility of measuring satisfaction with painmanagement in nursing home residents.

Table 2. Ratings of Overall Satisfaction and Satisfaction with Medical Therapy

RatingMean � Standard

Deviation Range Mode

Overall satisfaction with pain management 4.0 � 1.8 1–6 5Satisfaction with medication 3.9 � 1.8 1–6 5Frequency of desirable outcomes (0 � never; 5 � all of the time)

Staff make an effort to treat pain 2.3 � 2.1 0–5 0Staff ask about pain 1.4 � 1.8 0–5 0

Frequency of desirable outcomes from medication (0 � never; 5 � all of the time)Reduces pain to comfortable level 2.1 � 1.9 0–5 0Allows increased activity 1.1 � 1.6 0–4 0Relieves pain quickly 1.6 � 1.9 0–5 0Improves ability to walk* 0.9 � 1.5 0–5 0Allows resident to enjoy life more 1.8 � 2.0 0–5 0Improves sleep 2.3 � 2.0 0–5 0Improves mood 1.7 � 2.2 0–5 0

Frequency of undesirable outcomes from medication (0 � never; 5 � all of the time)Drowsiness 1.0 � 1.6 0–5 0Difficulty concentrating 0.6 � 1.4 0–5 0Dry mouth 2.0 � 2.1 0–5 0Nausea 0.3 � 1.0 0–5 0Constipation 0.8 � 1.5 0–5 0

*Residents who used a wheelchair or amputees who indicated they did not use a prosthesis were not asked this question.

Table 3. Correlates of Overall Satisfaction and Satisfaction with Medication

CharacteristicCorrelation with Overall

Satisfaction (Spearman Rho)

Characteristics associated with overall satisfaction (possible range)Current pain (0–7) �0.28*Pain in past week (1–7) �0.27*Staff efforts to assess pain (0–5) 0.25§

Staff efforts to treat pain (0–5) 0.51‡

Satisfaction with medication (0–5) 0.73‡

Geriatric Depression Scale score (0–15) �0.50‡

Characteristics associated with residents’ satisfaction with medication Perceived frequency of desirable outcomes (possible range)

Reduces pain to comfortable level (0–5) 0.57‡

Allows increased activity (0–5) 0.36*Relieves pain quickly (0–5) 0.66‡

Improves ability to walk (0–5) �0.01Allows resident to enjoy life more (0–5) 0.37†

Improves sleep (0–5) 0.44†

Improves mood (0–5) 0.46‡

Perceived frequency of undesirable outcomes (possible range)Confusion (0–5) �0.04Drowsiness (0–5) �0.09Difficulty concentrating (0–5) �0.33*Dry mouth (0–5) �0.003Nausea (0–5) �0.15Constipation (0–5) �0.13

*P � .05; †P � .01; ‡P � .001; §not significant: P � .051.

JAGS DECEMBER 2002–VOL. 50, NO. 12 SATISFACTION WITH PAIN MANAGEMENT 2033

First, these data suggest that nursing home residents’judgments of their satisfaction with pain management arestable over time. Moreover, these assessments appear to bestable even in residents with MMSE scores of 21 or less.These findings suggest that overall satisfaction with painmanagement, like satisfaction with other dimensions ofcare,17,18 can be assessed reliably in nursing home popula-tions. Second, overall satisfaction was inversely correlatedwith pain, albeit weakly. This finding is reassuring in light ofstudies in other settings that failed to find this relationship.8,9

Although these results are encouraging, other results sug-gest that assessments of satisfaction with pain managementshould be interpreted cautiously in nursing home populations.For instance, the finding that an assessment of satisfaction isstable over time raises questions about its responsiveness tochange. Specifically, if satisfaction is highly stable over time, itmay not be an optimal measure to evaluate treatment effectsor the outcomes of quality improvement interventions.

Another concerning finding is that, although pain sever-ity and satisfaction demonstrated an inverse correlation over-all, this correlation appears to exist only in residents withmore severe pain. That is, this evidence of satisfaction’s valid-ity as a measure was found only in residents with severe pain.This suggests that satisfaction may not provide a measure ofquality of care in residents with mild pain.

A third reason for concern lies in the significant in-verse relationship between pain satisfaction and the GDSscore, which persisted after adjusting for pain severity.This result is consistent with those of other studies thathave described a similar association between depressionand satisfaction with care.13,24–26 Because depression andpain both often go undetected and thus untreated in thenursing home setting, these findings underscore the impor-tance of screening for depression. Any assessments of sat-isfaction should be interpreted in light of the presence ofdepressive symptoms.

These results also suggest that measures of satisfactionmay not be appropriate for all residents. For instance, ap-proximately 34% of this nursing home population wasunable to participate in this study because of cognitive im-pairment. Therefore, it is not known whether reliable andvalid assessments of satisfaction are possible in these resi-dents. It is also possible that including the other questionsin this interview enhanced the validity of these satisfactionratings. In addition, six residents (9%) in this sample wereunable to rate their overall satisfaction with pain treat-ment. Further research is needed to better define residents’inability to answer this question and to find other qualitymeasures that can be used with residents who are too cog-nitively impaired to rate satisfaction.

This study’s results also provide mixed results aboutthe value of assessing residents’ satisfaction with painmedications. However, satisfaction with medication wascorrelated with commonly measured endpoints of paintreatment, such as speed of relief. These results indicatethat a single question that assesses satisfaction with painmedication could be sensitive to a variety of positive out-comes. However, only 90% of residents knew that theywere taking a medication for pain. This suggests that med-ication review and patient education should accompanyassessments of satisfaction. For example, staff who assessresidents’ satisfaction might remind them of medication

they are taking for pain. Similarly, nursing staff who passout medications have an opportunity to remind residentsof the indications for their pain medications.

It is also somewhat concerning that this study did notfind any correlations between residents’ satisfaction withmedication and the frequency of side effects that the resi-dent attributed to that medication. Residents who believedthat their pain medication caused frequent side effectswere nevertheless quite satisfied with it. Residents’ assess-ments of satisfaction with pain medication may be rela-tively insensitive to these effects, and staff could also in-quire about these directly.

Although these findings are potentially important, thisstudy had two limitations. First, it had a small sample sizeand may therefore have failed to find relationships be-tween variables reported here. It had adequate power todetect a relationship between overall satisfaction and painseverity, but other nonsignificant relationships should beinterpreted with caution. Second, the inclusion of a VAnursing home, whose residents were almost entirely men,made this sample somewhat atypical of other populationsof nursing home residents in whom pain and pain assess-ment have been studied.2,3 Nevertheless, the inclusion ofthis nursing home added substantial diversity, and there isno reason to believe that a different sex mix would haveinfluenced the associations reported here.

In summary, these results provide preliminary but im-portant data about the role that assessments of satisfactionmay play in ongoing efforts to improve the recognitionand treatment of pain in the nursing home. These resultssuggest that measures of overall pain satisfaction mayprove to be a valuable addition to care planning and maybe supplemented by assessments of satisfaction with medi-cation treatment, where appropriate. These assessmentshave the potential to make care plans more resident cen-tered, which has been advocated as a goal of long-termcare.7,27 However, further research is needed to define theusefulness of these measures in larger populations.

ACKNOWLEDGMENTSThe authors are grateful for the support of Drs. Leslie Car-son and Charles Spencer and the staff and residents whocontributed to this project.

REFERENCES1. Ferrell BA, Ferrell BR, Osterweil D. Pain in the nursing home. J Am Geriatr

Soc 1990;38:409–414.2. Ferrell BA, Ferrell BR, Rivera L. Pain in cognitively impaired nursing home

patients. J Pain Sympt Manage 1995;10:591–598.3. Weiner D, Peterson B, Ladd K et al. Pain in nursing home residents. An ex-

ploration of prevalence, staff perspectives, and practical aspects of manage-ment. Clin J Pain 1999;15:92–101.

4. Parmelee PA, Smith B, Katz IR. Pain complaints and cognitive status amongelderly institution residents. J Am Geriatr Soc 1993;41:517–522.

5. Skevington SM. Investigating the relationship between pain and discomfortand quality of life, using the WHOQOL. Pain 1998;76:395–406.

6. Won A, Lapane K, Gambassi G et al. Correlates and management of nonma-lignant pain in the nursing home. J Am Geriatr Soc 1999;47:936–942.

7. Joint Commission for the Accreditation of Health Care Organizations.2002–2003 Comprehensive Accreditation Manual for Long Term Care.Oakbrook Terrace, IL: Joint Commission Resources Inc., 2002.

8. Ward SE, Gordon DB. Patient satisfaction and pain severity as outcomes inpain management: A longitudinal view of one setting’s experience. J PainSympt Manage 1996;11:242–251.

9. Miaskowski C, Nichols R, Brody R et al. Assessment of patient satisfaction

2034 CASARETT ET AL. DECEMBER 2002–VOL. 50, NO. 12 JAGS

utilizing the American Pain Society’s Quality Assurance Standards on acuteand cancer-related pain. J Pain Sympt Manage 1994;9:5–11.

10. Stahmer SA, Shofer FS, Marino A et al. Do quantitative changes in pain intensitycorrelate with pain relief and satisfaction? Acad Emerg Med 1998;5:851–857.

11. AGS Panel on Persistent Pain in Older Persons. The management of persis-tent pain in older persons. J Am Geriatr Soc 2002;50:S205–S224.

12. Magaziner J, German P, Zimmerman S et al. The prevalence of dementia in astatewide sample of new nursing home admissions aged 65 and older: Diag-nosis by expert panel. Gerontologist 2000;40:663–672.

13. Desbiens NA, Wu AW, Broste SK et al. Pain and satisfaction with pain con-trol in seriously ill hospitalized adults: Findings from the SUPPORT researchinvestigations. For the SUPPORT investigators. Study to Understand Prog-noses and Preferences for Outcomes and Risks of Treatment. Crit Care Med1996;24:1953–1961.

14. McCracken LM, Klock PA, Mingay DJ et al. Assessment of satisfaction withtreatment for chronic pain. J Pain Sympt Manage 1997;14:292–299.

15. Cohen J. Statistical Power Analysis for the Social Sciences, 2nd Ed. Hillsdale,NJ: Lawrence Erlbaum Associates, 1988.

16. Gracely RH, Dubner R. Reliability and validity of verbal descriptor scales ofpainfulness. Pain 1987;29:175–185.

17. Simmons SF, Schnelle JF. Strategies to measure nursing home residents’ satis-faction and preferences related to incontinence and mobility care: Implica-tions for evaluating intervention effects. Gerontologist 1999;39:345–355.

18. Ryden MB, Gross CR, Savik K et al. Development of a measure of residentsatisfaction with the nursing home. Res Nurs Health 2000;23:237–245.

19. Folstein MF, Folstein F, McHugh PR. “Mini-mental state”. A practicalmethod for grading the cognitive state of patients for the clinician. J PsycholRes 1975;12:189–198.

20. Lesher EL, Berryhill JS. Validation of the Geriatric Depression Scale—ShortForm among inpatients. J Clin Psychol 1994;50:256–260.

21. Alden D, Austin C, Sturgeon R. A correlation between the Geriatric Depres-sion Scale long and short forms. J Gerontol 1989;44:124–125.

22. Lyness J, Noel T, Cox C et al. Screening for depression in elderly primarycare patients. A comparison of the Center for Epidemiologic Studies—De-pression Scale and the Geriatric Depression Scale. Arch Intern Med 1997;157:449–454.

23. Prigerson HG, Frank E, Kasl SV. Complicated grief and bereavement-relateddepression as distinct disorders: Preliminary empirical validation in elderlybereaved spouses. Am J Psychol 1995;152:22–30.

24. Haviland MG, Dial TH, McGhee WH et al. Depression and satisfaction withhealth plans. Psychiatr Serv 2001;52:279.

25. Nolen-Hoeksema S, Larson J, Bishop M. Predictors of family members’ satis-faction with hospice. Hospital J 2000;15:29–48.

26. Scherer MJ, Cushman LA. Predicting satisfaction with assistive technologyfor a sample of adults with new spinal cord injuries. Psychol Rep 2000;87:981–987.

27. Blair C. Residents who make decisions reveal healthier, happier attitudes. JLong Term Care Adm 1994;22:37–39.

Appendix 1. Sample Item from Instrument

Overall satisfaction

Overall, how satisfied are you with the way that your pain is being treated?�1—Very dissatisfied�2—Dissatisfied�3—Slightly dissatisfied�4—Slightly satisfied�5—Satisfied�6—Very satisfied