Symptoms in patients with lung carcinoma

Distinguishing distress from intensity

Authors


Abstract

BACKGROUND

The patient perspective on distress associated with lung carcinoma is important, yet understudied. Previous research on symptom experience generally had not differentiated the dimension symptom intensity/frequency from which symptoms are associated with most distress. The objective of the current study was to determine whether patterns of symptom intensity were similar to patterns of symptom distress, whether patterns were consistent at different time points, whether patterns varied by subgroups, and whether high symptom intensity was equivalent to distress.

METHODS

Four hundred adults who were newly diagnosed with inoperable lung carcinoma completed a measure of symptom intensity/frequency and a new measure of distress associated with symptoms at six time points during the first year after diagnosis. These data were supplemented by field notes by research nurses and by less structured, qualitative interviews.

RESULTS

The mean ranking of distress in the total group and in all subgroups remained constant at all time points, with breathing, pain, and fatigue associated with the most distress. In contrast, the pattern of mean rank order of symptom intensity showed little consistency; however, fatigue had the highest intensity scores at all time points.

CONCLUSIONS

The current data challenged the uncritical use of summated scores of different symptom items in the context of lung carcinoma. Breathing and pain appeared to function as icons representing threats associated with lung carcinoma, with distress described as related to the past and the present and to expectations for the future. One of the most promising implications of these data was in fostering a preventive paradigm for symptom palliation. Cancer 2005. © 2005 American Cancer Society.

Patients with lung carcinoma suffer severe symptomatology compared with other groups of cancer patients.1, 2 High mortality, with a low percentage of lung carcinoma patients surviving the first year after diagnosis, means that issues related to palliation, death, and dying often become relevant at diagnosis. Despite the vulnerable situation for these patients, published findings concerning symptom experience and quality of life (QL) remain relatively sparse compared with data for other cancer sites,2, 3 although their value is emphasized increasingly as a basis for planning and evaluating care.4, 5 In a secondary analysis of three longitudinal studies on symptom distress in cancer patients, Cooley et al.6 noted that the vast majority of studies published to date have been cross-sectional, with a lack of data regarding changes in symptom experience over time.

In addition to the dearth of data, lack of both theoretical clarity and clarity in dimensions of measurement create difficulties in interpreting and translating knowledge into clinical practice. Hollen and colleagues5, 7 have added a degree of precision in addressing the relation between the broad concept of QL and symptomatology as a key dimension of QL. Hollen et al.7 noted that individuals may weight specific aspects of QL differentially as more or less important, which, thus, may affect differentially the perception of distress. Despite this, previous theoretical work and studies of symptoms in cancer patients largely have focused on the current intensity or frequency of symptoms, although this rarely is differentiated from symptom distress.8–13 The question of how distressing patients perceive different symptoms to be, irrespective of their current levels of occurrence, has not been addressed specifically, although several studies have suggested that the dimensions of frequency/intensity of symptoms are not equivalent to distress.14–16 One explanation may be that distress (i.e., how disturbing or troubling a symptom is perceived to be) refers to meanings that the illness holds for an individual and that these meanings are relative to one's life. This anthropologic, interpretative, or ‘meaning-centred’ approach17 acknowledges that symptoms are more than a direct reflection of disease. Understanding what are the most distressing symptoms from the perspectives of patients, thus, may provide a new orienting framework to guide interventions.

The current study was designed to address four questions: 1) Are patterns of current symptom intensity similar to patterns of distress associated with symptoms in a naturally occurring population of newly diagnosed patients with inoperable lung carcinoma? 2) Are patterns of symptom intensity and distress consistent across time points? 3) Are there differences in patterns of intensity and distress in subgroups of lung carcinoma patients based on selected demographic and disease-related variables? 4) Is high symptom intensity equivalent with symptom distress? The overall objective of this study was to provide an empiric basis to ensure that patient perspectives are accommodated in efforts to improve symptom management.

MATERIALS AND METHODS

This study consisted of a repeated-measures, descriptive survey in a naturally occurring sample of 400 adults who were recruited consecutively close to the time that they were diagnosed (mean, 31 days; median, 23 days) with primary, inoperable lung carcinoma through the lung medicine departments of 2 university hospitals in Stockholm between April 1998 and October 2002. These hospitals are the primary centers for the nonsurgical treatment of lung cancer in the Stockholm region.

In addition to providing basic demographic and medical/treatment information, patients completed several self-report instruments at 6 time points: close to diagnosis and before the initiation of treatment (T1), 2 weeks after T1 (T2), 1 month after T1 (T3), 3 months after T1 (T4), 6 months after T1 (T5), and 1 year after T1 (T6). Patients who chose to participate in the study after they had started treatment were entered at T2. The time points were selected to provide assessments both on treatment and off treatment, to capture changes that may occur during the process of treatment, and to provide information regarding potential target symptoms for future home care interventions.

The project was initiated after receiving approval from the relevant medical research ethics committee (Karolinska Institutet registration number [KI dnr] 97-258, appendices 990503, 010220, 010731, 021212). Staff at the lung medicine departments were asked to give patients a first information letter, with a copy also sent by the research team to the patient's home. A research nurse contacted the patient several days later to answer any questions and to obtain informed consent for study participation. All data were collected in the presence of a research nurse at a venue determined by the patient, including the patient's home, hospital clinic, and inpatient hospital ward. Efforts were made for the same research nurse to conduct all interviews with the same patient. At the conclusion of each interview, the participant was offered a 50-sek (approx €5.58 or $7.30) gift certificate donated by COOP Sweden, which is a national federation of cooperative stores.

The data for this article were composed of an adapted version of the McCorkle and Young Symptom Distress Scale (SDS)18, 19 and the Thurstone Scale of Symptom Distress-Lung Cancer (TSSD-LC),14 which were completed at all time-points. The quantitative data were supplemented with field documentation by the research nurses and, in some patients (n = 95), with unstructured, audiotaped interviews.

The Adapted SDS

The SDS is a well known self-report measure that is used commonly in nontrial studies of symptoms in patients with lung carcinoma.3, 18, 19 Although the definition of symptom distress used when developing the SDS was “the degree of discomfort from the specific symptom being experienced as perceived by the patient,” it is argued here that the SDS investigates symptom occurrence rather than distress, because patients are asked primarily to rate intensity and frequency. The Swedish visual analog scale version of this study uses items that are included both in the original version and a later version that was revised to reflect symptoms specific to lung carcinoma,19–21 consisting of 15 items that explicitly measure the intensity of 10 symptoms (fatigue, insomnia, appetite, mobility, concentration, mood, outlook, appearance, pain, and nausea), with frequency of pain and nausea also assessed. Three symptoms (bowel function, cough, and breathing) are formulated ambiguously as assessing either intensity or ‘troublesomeness.’ The SDS was presented to patients on 15 cards, 1 for each item, that assessed symptom occurrence at the present time. The items are scored by measuring distance from the beginning of the scale to the mark made by the patient across the visual analog scale line, thus yielding a score between 0 and 10 for each item, with higher scores indicating greater perceived problem. The Cronbach α for the SDS used here varied between 0.84 and 0.88 across the time points, comparing well with the value of 0.81 that resulted from the use of an earlier Swedish Likert version22 and reported Cronbach α scores between 0.70 and 0.92 in 47 studies using an English versions of the 13-item SDS.18

The TSSD-LC

A Thurstone scale of pairwise comparisons of symptoms that were perceived as distressing was developed and pilot tested14 for use in the current study. To reduce patient fatigue, nine commonly occurring symptoms (outlook, breathing, pain, insomnia, cough, bowel function, appetite, fatigue, and appearance) were selected from the SDS based on an earlier Canadian study23 and were reconfirmed as appropriate in a Swedish pilot study.14 The 9 symptoms were arranged in pairs of every possible subset of 2, yielding 36 pairs. The symptom pairs were presented to the patient in a set with one symptom pair per card, using the Ross matrix of optimal ordering24 to avoid selection bias. Participants were asked to select the symptom associated with the most distress, irrespective of the occurrence of that symptom at the moment. Because only two words were presented on each card, readability was not an issue. The presence of the research nurse allowed for clarification of instructions when necessary.

Thurstone postulated that, given a set of stimuli (i.e., nine symptoms), each would possess some attribute (in this instance, distress) in various but unknown degrees. For each of the stimuli and among all participants, it is assumed that a preference will exist and that, for each symptom, the perceived distress will be distributed normally around that symptom's modal response, which forms the scale values to rank the symptoms. The number of patients who prefer Symptom 1 to Symptom 2, Symptom 1 to Symptom 3, etc., yields preference frequencies, which are divided by the number of patients responding, to calculate the percentage of the sample that prefers each symptom to every other symptom. These proportions are then translated into standard normal scores. The more respondents who select one symptom in a pair over the other, the greater the preference for that symptom and the greater its score; thus, higher scores indicate more distress associated with that symptom.25

Time in disease trajectory was the primary variable used in planning sample size. A difference of 10% in the average proportion of times an item is chosen in 1 symptom profile compared with another is deemed clinically relevant with 200 patients per group sufficient to detect statistical differences with 80% power (P = 0.05). Symptom profiles for equality of proportions in subgroups are compared using a Bonferroni correction. The test statistic differs from more common Z-statistic testing, in that it incorporates a multiplier of (square root)k-1, because the proportions are an average of k-1 individual proportions rather than singular proportions themselves.

Two measures of reliability were applied to the generated data. The internal consistency of each individual's comparative judgments was identified by computing the number of circular triads or inconsistencies present in each patient's comparisons. Between 96.7% and 98.7% of all patients were consistent in their preferences at each time point; those patients with ≥ 6 missing values were excluded from the TSSD-LC analysis (28 interviews from 16 different patients, of 1428 interviews from 400 patients, were excluded for this reason). The Gulliksen and Tukey index of scalability, which is an estimate of reliability, was 0.95–0.98 across time points.

The Kendall coefficient of agreement was used to measure the degree of consistency among participants. The closer this value is to 1.00, the closer the patients are to complete agreement about the relative distress from the different symptoms. The Kendall coefficients in this study ranged from 0.24 to 0.31, indicating a low-moderate degree of agreement with respect to symptoms associated with most distress in patients with inoperable lung carcinoma.

RESULTS

Sample

In total, 910 patients were invited to participate, and there were 400 final participants. Reasons for nonparticipation are outlined in Table 1. Most nonparticipants did not participate because they declined without a specified reason or because they were too tired. The study participants were younger (P < 0.001) and survived longer (mean survival, 320 days; P = 0.001) than nonparticipants (mean survival, 263 days). The study participants also differed from population data from newly diagnosed patients with inoperable lung carcinoma during the same period in the Stockholm-Gotland Cancer Registry. The cancer registry population was older (n = 2375 patients; mean age, 68.8 yrs) and had shorter survivals (n = 1975 patients; mean survival, 178 days) than study participants, which may have been because of the registration of patients who were diagnosed in nonuniversity hospital settings.

Table 1. Attrition and Reasons for Nonparticipation as Reported by Patients
VariableNo. of patients%
Patients invited to participate910100
Decreased before data collection313
Could not be reached for response233
Declining participation45650
No reason specified14816
Too tired to participate12614
Lack of time505
Poor physical condition415
Critical of the study182
Cognitive problems142
Other reasons596
No. of participants40044

Table 2 shows that the sample was divided relatively evenly between men and women, but men were significantly older than the participating women. Men were more likely to be married, whereas women more often were divorced and widowed. There were no significant differences between men and women with regard to educational level, type of lung carcinoma, or stage of lung carcinoma. There was a significant difference in treatment received between men and women, with 84% of women and 72% men receiving some form of chemotherapy (P = 0.006). Fourteen percent of men and 6% of women received no oncologic treatment.

Table 2. Baseline Demographic and Clinical Data
VariableAll patientsMale patientsFemale patientsP value
No.%No.%No.%
  • SD: standard deviation; SCLC: small cell lung carcinoma; CT: chemotherapy; RT: radiotherapy; T1: close to diagnosis and prior to the initiation of treatment; T2: 2 weeks after T1.

  • a

    Chi-square computation was based only on valid cases: degrees of freedom (df) = 3.

  • b

    Chi-square computation was based only on valid cases: df = 2.

  • c

    Chi-square computation was based only on valid cases: df = 1.

  • d

    Chi-square computation was based only on valid cases: df = 5.

  • e

    Chi-square computation was based only on valid cases: df = 4.

No. of patients4001002095219148 
Age in yrs       
 Median63 67 60  
 Mean64.5 66.2 62.6 0.001
 SD10.5 10.7 9.9  
Marital status       
 Married25363154749952 
 Divorced/separated641623114122 
 Widow/widower47121473317 
 Unmarried328188147 
 Missing410042< 0.001a
Educational level       
 > High school1283275365328 
 High school or equivalent (12 yrs)751931154423 
 < High school (9 yrs)1854697468846 
 Missing12363630.059b
Type of lung cancer       
 SCLC561427132915 
 Non-SCLC339851798616084 
 Missing5131210.524c
Stage       
 Stage I1137342 
 Stage II205105105 
 Stage IIIA359178189 
 Stage IIIB781943213518 
 Stage IV1644181398344 
 Unclassified tumor23617863 
 Missing6917341635180.318d
Treatment received       
 CT alone19549904310555 
 RT alone491230141910 
 Concomitant CT and RT511329142212 
 Both CT and RT641631153317 
 No treatment41102914126 
 Missing      0.028e
Accrual       
 T1245611426810354 
 T215539673288460.004

Patterns of Symptom Distress

Figure 1 illustrates the relative ranking of symptoms on the dimension ‘distress’ across 6 time points according to the TSSD-LC. There is a consistent pattern with breathing, pain, and fatigue, respectively, as the items associated with most distress in the same order across all time points. Appearance always was ranked lowest; that is, it was associated with the least distress. The remaining symptoms were differentiated less and tended to be clustered closely together in the middle of the scale. Because of the complex nature of missing data in this severely ill group of patients, we compared the data arrayed in Figure 1 with a subsample of the same 164 patients at T2, T4, and T5 and found the same pattern of symptoms ranked highest and lowest on the dimension of ‘distress’ (data not shown). Consequently, further analyses were conducted with the entire sample accrued at each time point.

Figure 1.

The relative ranking of symptoms on the dimension “Distress” according to the Thurstone Scale of Symptom Distress-Lung Cancer. T1: close to diagnosis and prior to initiation of treatment; T2: 2 weeks after T1; T3: 1 month after T1; T4: 3 months after T1; T5: 6 months after T1; T6: 1 year after T1.

The overall pattern of the three top-ranking symptoms did not change across subgroups defined by age, gender, and type of lung carcinoma, with one exception. Breathing, pain, and cough were the top-ranked symptoms among the 10 patients with SCLC at T6. Although the pattern of highest ranked symptoms was consistent otherwise, there were some noteworthy differences in the proportions of subgroups that responded in this manner at particular time points. No differences in subgroups were found with regard to breathing, but the average preferred proportion of older patients (age younger than 64 yrs vs. age 64 yrs and older) who ranked distress from fatigue was higher than for younger patients at T3 and was higher for men than for women at T4. Women at T1 and younger patients at T3 ranked distress from pain higher. Finally, patients with SCLC at T1 ranked pain higher than patients with NSCLC.

Patterns of Symptom Intensity

The SDS data contrasted markedly with the TSSD-LC data in terms of consistency, as shown in Table 3. The mean SDS scores for fatigue consistently were the highest, ranging from 5.06 at T1 to 4.18 at T6, and they were significantly higher than the scores for the next most intense symptom (P < 0.05). No other consistent patterns were found in the rank ordering of the other symptoms. Subgroup analyses using the same variables described above (age, gender, and type of lung cancer) indicated that younger patients and women reported significantly higher intensity problems related to outlook at all time points, as did patients with SCLC at T5. Appearance problems were greater for women than for men at T2, T3, and T5; for younger patients at T5; and for those with SCLC at T3, T4, and T6. Higher levels of fatigue were reported by patients with SCLC at T4–T6 and by women at T6. Patients with NSCLC reported higher levels of cough at T3 and T4 than patients with SCLC. At T6, younger patients reported higher levels of bowel problems than their older counterparts, and women reported higher levels of pain and appetite disturbances than men.

Table 3. Symptom Distress Scale Mean Scoresa
T1 (n = 228)T2 (n = 325)T3 (n = 313)T4 (n = 260)T5 (n = 193)T6 (n = 81)
SymptomMean scoreSymptomMean scoreSymptomMean scoreSymptomMean scoreSymptomMean scoreSymptomMean score
  • T1: close to diagnosis and prior to initiation of treatment; T2: 2 weeks after T1; T3: 1 month after T1; T4: 3 months after T1; T5: 6 months after T1; T6: 1 year after T1.

  • a

    Symptom Distress Scale data are reported only for the symptom items that also appear on the Thurstone Scale of Symptom Distress-Lung Cancer.

Fatigue5.1Fatigue4.7Fatigue4.4Fatigue4.7Fatigue4.3Fatigue4.2
Breathing3.6Appetite3.3Appetite2.9Appearance3.1Outlook3.1Appetite3.2
Outlook3.4Insomnia3.2Insomnia2.8Outlook2.9Pain3.0Pain3.1
Cough3.4Breathing3.1Outlook2.8Breathing2.9Appetite2.8Breathing3.1
Appetite3.3Outlook3.1Breathing2.8Appetite2.8Appearance2.6Outlook3.0
Insomnia3.2Cough2.8Pain2.6Pain2.5Insomnia2.6Appearance3.0
Pain2.6Bowel2.7Appearance2.6Bowel2.4Breathing2.6Cough2.9
Appearance2.1Pain2.6Bowel2.4Cough2.4Cough2.3Insomnia2.4
Bowel2.1Appearance2.4Cough2.3Insomnia2.3Bowel2.1Bowel2.2

Is High-Level Symptom Intensity Equivalent to Symptom Distress?

To explore whether high-level symptom intensity may be equated with distress, further analysis was conducted on the symptoms ranked as the top three on the TSSD-LC. Table 4 illustrates the proportion of patients who scored in the top one-third of the visual analogue scale for intensity of symptoms along with the proportions of patients who ranked the same symptom as highest on the TSSD-LC. Particularly with respect to breathing and fatigue, the proportions of patients who reported high symptom occurrence differed from the proportions that reported the same symptom as that associated with most distress. Between 9% and 22% of patients scored on the top one-third of severity in problems with breathing over time, whereas between 38% and 50% of patients reported that this symptom was associated with the most distress. An opposite pattern was evident for fatigue, with 19–37% of patients reporting high levels of fatigue, but a notably lower proportion of patients associated this symptom with the most distress. A difference in dimensions still was evident but was less marked with respect to pain.

Table 4. The Percentage of Patients who Ranked Breathing, Pain, or Fatigue Highest on the Thurstone Scale of Symptom Distress-Lung Cancer Compared with the Percentage of Patients with High-Level Intensity on the Symptom Distress Scale
Time pointBreathing (% of responses)Pain (% of responses)Fatigue (% of responses)
TSSD-LC (highest score)SDS score ≥ 6.7TSSD-LC (highest score)SDS score ≥ 6.7TSSD-LC (highest score)SDS score ≥ 6.7
  1. TSSD-LC: Thurstone Scale of Symptom Distress-Lung Cancer; SDS: Symptom Distress Scale; T1: close to diagnosis and prior to initiation of treatment; T2: 2 weeks after T1; T3: 1 month after T1; T4: 3 months after T1; T5: 6 months after T1; T6: 1 year after T1.

T138.021.614.510.613.636.9
T238.715.720.89.111.228.2
T339.711.520.010.310.723.5
T440.612.419.77.011.027.8
T544.09.419.013.48.225.5
T650.014.818.813.96.318.5

In an attempt to illustrate which symptoms ranked highest according to distress, as represented from the patients' perspective, spontaneous comments offered by participants and documented as quotes in field notes by the research nurses were reviewed for manifest content.26 Temporal variations were striking, and some typical responses are presented in Table 5. Breathing and pain were described with reference to the past, the present, and expectations for the future; while fatigue, with few exceptions, was described in the present tense.

Table 5. Examples of Participant Descriptions of Symptoms Ranked Highest on the Thurstone Scale of Symptom Distress-Lung Cancer
Breathing
 “Overshadows everything else”
 “The main thing is that you can breathe”
 “…Something I've always been afraid of”
 “You have no future if you can't breathe”
 “I'd have a completely different life if they could do something about my breathing”
 “No problem now, but can't stand the thought”
 “Breathing has always been difficult for me…always thought that I'll die from my breathing problems”
Pain
 “The worst that could happen”
 “Pain is something you can put up with”
 “They can help you with pain”
 “Uncertainty about what the pain is”
 “Pain gets the upper hand”
 “I'm very afraid of having pain”
 “Some people have terrible, terrible pain when they die from this disease…I had an in-law who died last month”
Fatigue
 “I'm tired of being tired”
 “My body's weak, have no strength in my legs”
 “Everything gets put aside”
 “Shut in”
 “Boring and depressing, being tired”
 “Fatigue—that would get me down altogether”
 “Energy leakage”
 “My body feels used ”

DISCUSSION

The most striking finding of this study was that the dimension of symptom intensity and the dimension of symptom distress were not equivalent. There also was remarkable consistency in the TSSD-LC data on symptom distress over time in all subgroups examined, whereas patterns of intensity were highly variable. The 3 symptoms—breathing, pain, and fatigue—were associated with the most distress in this sample of 400 patients with inoperable lung carcinoma.

It should be noted, however, that the sample in this study represents the ‘healthiest’ of this very ill patient group, because participants were younger and had longer survivals than both nonparticipants and the population of patients with inoperable lung carcinoma registered in the same geographic area. Recruitment and participation in this study compared favorably with other studies, considering the well acknowledged difficulties in researching patients with lung carcinoma, debilitated patients, and women with lung carcinoma.3, 27–30 The low levels of attrition and of missing and inconsistent data support the research team's impression that patients were not overburdened by the instrument battery.

‘Breathing’ and ‘pain’ may function as icons that represent threats associated with lung carcinoma. The symbolic value of these symptoms, therefore, may cause their strong association with distress, even in times of relatively low symptom occurrence. Breathing may be considered a symbol of life itself, with dyspnea perceived as particularly threatening: Without the ability to breathe, one will no longer be able to live.31, 32 Pain often is considered an icon of a cancer in general. Pain has been identified as one of the symptoms most feared and associated with cancer and death both by patients and in the general population.33, 34

Fatigue was the most intense symptom on the SDS. The study data suggest that breathing and pain are interpreted as more threatening than the relatively intense fatigue experienced by these patients. It is conceivable that fatigue may have greater consequences for daily life than lower levels of pain and dyspnea35 but that ‘distress’ is a concept that has implications beyond the current moment.

The differences in time frame for each measure must be considered in interpreting the current results. The SDS reports subjective symptomatology at the time of questionnaire response, which is the case with most symptom measurements. The TSSD-LC, conversely, was designed in accordance with a ‘meaning-centred approach’ to investigate a broader time frame. Thus, the patient was asked to select the symptom he or she would find most distressing, irrespective of current symptom occurrence. This allows the TSSD-LC to address a time frame that incorporates both past and current experiences as well as fears or expectations for the future. It is noteworthy that temporality rarely has been addressed with regard to symptom distress. Even in the recent report by Henly et al. in which they provided an overview of the notion of time in symptom experiences,36 the role of future hopes and expectations about symptom experiences was not discussed.

These data challenge the uncritical use of summated scores of different symptom items associated with highly variable levels of distress. Given the wide range in scale scores on the TSSD-LC between the top and bottom ranked items, it seems fallacious to weight them equally on a summated scale. Hinds et al.37 pointed out the negative implications for appropriate patient intervention when relying on summated scores rather than individual assessments of each symptom alone. However, examining each symptom individually, as we did in the current study with the SDS, may exacerbate known weaknesses in comparability both between individuals and between items that exist in visual analogue scales.38, 39 In these SDS data, with the exception of fatigue, only minor differences were found in mean scores for intensity. Although the SDS can provide useful information for symptom treatment, in situations such as this, with close proximity among most SDS mean values, data that reflect patient prioritizations provide a compelling basis for clinical intervention.

To address limitations in measures like the SDS, other instruments, most notably, the Memorial Symptom Assessment Scale (MSAS), have been developed in the past decade to rank the dimension of distress associated with symptoms. Whereas the ranking of distressing symptoms derived from the TSSD-LC is based on a forced-choice comparison between pairs of symptoms, the ranking of distress associated with symptoms on the MSAS is obtained using a Likert-type format arrayed side-by-side with an evaluation of symptom severity and frequency using the same format. It is similar to the SDS, in that there is no forced comparison between symptoms.

The MSAS combines a set group of frequently occurring and often distressing symptoms into what they call a ‘global distress index’ (GDI). Apart from the GDI, MSAS data on distress often are not reported separately; when they are reported, results differ from those presented here, in that the proportion of patients describing frequency/intensity of a symptom as relatively high exceeded the proportion reporting the same symptom as distressing.9 The difference in MSAS results and these data may be a reflection of the MSAS focus on current symptomatology versus the broader time frame captured by the TSSD-LC. An alternative explanation may be differences in symptom experience in patients with lung cancer compared with cancer populations studied with the MSAS.

The TSSD-LC demonstrated promising psychometric properties in the current study. Despite the severity of illness of the participating patients, they were able to complete the multiple comparisons of symptoms with few inconsistencies in their paired comparisons. Thus, these data compare favorably with the results obtained in less ill cancer populations using Thurstone scaling for other purposes.40, 41 The index of scalability was high, suggesting that, although a forced-choice format was used, the data reflect a viable dimension. The low-moderate Kendall coefficient of agreement suggests a need for future individual assessment to develop personalized ‘symptom ladders’ to facilitate communication between patients and health providers. The recent availability of programming for Thurstone analysis (available at URL: http://www.umanitoba.ca/centres/mchp/concept/thurstone [accessed September 13, 2005]) makes this type of analysis more accessible to other investigators, and the implementation of a touch-screen computer format for the paired-comparison task should enhance further patient acceptability of the measure.42

One of the most promising implications of these data may be in fostering a preventive paradigm in palliative care, which traditionally has focused on managing symptoms after they arise. In the current study, it appeared that distress was related to fears for the future and to past and present experiences. In line with the recent redefinition of palliative care by the World Health Organization,43 attention paid to preventing symptoms feared by patients, by virtue of past experiences or expectations of the future that awaits them, is well in accordance with a public health perspective on palliative care.44 The novel approach to measuring distress associated with symptoms from lung carcinoma reported here has the potential to bridge the existing gap between individuals and populations, because it can be applied on both levels. The knowledge that breathing, pain, and fatigue are the symptoms that patients with inoperable lung carcinoma associate with most distress provides a strong, empiric basis for preventive intervention and clinical education.

Ancillary