What does the answer mean? A qualitative study of how palliative cancer patients interpret and respond to the Edmonton Symptom Assessment System

Introduction

Adequate symptom assessment is of the utmost importance for optimal symptom management and maintenance of quality of life (QoL), which are the goals of palliative treatment and care. ¹ Studies have shown that lack of systematic symptom assessment is one of the reasons for sub-optimal symptom management.^2–4 Because symptoms are subjective by nature, it is generally agreed that the patient is the most valuable source of information.^5,6 The Edmonton Symptom Assessment System (ESAS) is a widely used and well-known self-reporting tool for assessment of symptoms in palliative care. ⁷ The ESAS was developed in order to monitor the most frequent symptoms in advanced cancer with minimal patient burden. ⁶ The tool is easy and expedient to use and can be used for day-to-day monitoring of treatment effect and change in symptoms, ⁶ all of which are important factors in advanced cancer.

The ESAS has been validated in several clinical and cultural settings, and has been compared with different symptom assessment tools, both disease and symptom specific.^7–10 Despite the extensive use and validation of the ESAS, recent research has questioned how well it works as a self-reporting tool. In a Canadian study, nurses were concerned about how well the patients understood the terminology of the tool. ¹¹ It has also been shown that patients tended to score some symptoms in the opposite direction of what they actually meant. For example, they answered 0 meaning ‘not tired’ when they actually meant the other extreme: 10 ‘worst possible tiredness’. ¹² Furthermore, most validation studies of the ESAS have primarily focused on obtaining quantitative convergence proofs of validity by comparing relatively similar tools. Qualitative studies of the patients’ perspective of self-reporting tools are less common.^13,14 However, the validity of an instrument relies on a common understanding of the meaning of the questions and the response categories. ¹⁵ Findings from psychology and sociology have shown that people often interpret questions in different and unexpected ways, compared to what was intended in the development process of the tool.^14–16

When using a self-reporting tool, potential sources of error are related to what happens during the process of completing the questionnaire.^15,16 The wording, the order of the questions and the response categories may influence the answers.^14–16 In order to understand how people interpret and respond to questionnaires, knowledge about the cognitive processes that take place when a person responds to the questions is required. The cognitive processes are: (1) comprehension and interpretation of the question; (2) retrieval of information from memory; (3) forming a judgment; and (4) editing the answers (self-reporting). ¹⁶ When examining these processes, it has been recommended that instead of asking whether a person understands a question, one should ask what the question means to the person. ¹⁶ This is because those who misunderstand are often not aware that they do so, which in turn makes the misunderstandings difficult to detect. ¹⁶ If the health care providers have insufficient knowledge about how the patients interpret and complete the ESAS, this may result in sub-optimal symptom management.

We have only been able to identify one study that has examined the ESAS from the patients’ perspective. Watanabe et al. ¹⁷ used the ‘think aloud’ method and found that the symptom ratings were influenced by several factors, such as symptom characteristics, temporal changes, previous symptom experiences and individual perceptions. They also found that many patients had difficulties interpreting some of the symptoms and using the numerical scales. For example, some patients had difficulties rating anxiety and depression and understanding the term well-being. The authors called for a replication of their study in other populations in order to provide further information about how the ESAS functions as an assessment tool. Thus, the present study was initiated to examine how palliative cancer patients interpret and respond to the ESAS by means of cognitive interviewing.

The following research questions were examined:

Material and methods

Results

Twenty-two patients were regarded as eligible over a period of four months, whereas 10 patients declined participation (Figure 1). One interview was not included in the analyses because the informant was cognitively impaired during the interview. Thus the sample consisted of 11 patients: three women and eight men, with mixed diagnoses, median age 65 (Table 1). Table 2 shows the mean intensity scores of the ESAS. As can be inferred from the table, the highest scores were found with tiredness (6.3) and oral dryness (5.7) and the lowest scores were found with nausea (2.2).

OPEN IN VIEWER

Figure 1.

Patient recruitment flowchart.

OPEN IN VIEWER

Table 1.

Patient characteristics

		n = 11
Gender	Female	3
	Male	8
Primary cancer	Pancreas	4
diagnosis	Gastrointestinal	2
	Genitourinary	4
	Head and neck	1
Age	Median 65
	Range 34–95

OPEN IN VIEWER

Table 2.

ESAS symptom intensity score (0–10)

		Mean	Range
Symptoms	Pain at rest	2.8	0–9
	Pain at movement	3.3	0–10
	Tiredness	6.3	1–10
	Nausea	2.2	0–8
	Shortness of breath	2.8	0–9
	Oral dryness	5.7	1–8
	Appetite	4.6	0–10
	Anxiety/uneasiness	2.5	0–8
	Depression/sadness	2.6	0–8
	‘Overall, how are you  feeling today?’	5.2	2–8

The themes that emerged during the analyses are presented with regard to the research questions: interpretation of the symptoms, influence of the response format and the influence of previous experience with the ESAS. In Table 3 a summary of the themes is presented.

OPEN IN VIEWER

Table 3.

A summary of the themes from the thematic analysis

1. Interpretation of the symptoms
• Comprehension of the symptoms
• The ‘well-being’ question in relation to other symptoms
2. Influence of the response format
• Misunderstandings
• Meaning of scores
• Contextual factors influencing the response given
• Meaning of the anchoring term ‘worst possible’
• Time frame
3. Influence of previous experience with the ESAS
• Information
• Feedback

Discussion

In this qualitative study the aim was to examine how a sample of palliative care cancer patients interpreted and responded to the questions in the ESAS. The findings showed that using a self-reporting tool, such as the ESAS, involves possible sources of error in relation to terminology and scoring. Errors occurred regarding the interpretation of the symptoms and both the response format and context factors influenced the judgement process during completion. Thus, patients’ interpretation should always be considered and health care providers should strive to minimize these types of errors by going through the form with the patients.

If a patient does not comprehend the questions as intended, the data may be neither valid nor reliable. The analyses revealed that both the symptoms anxiety and depression were most often subject to difficulties in interpretation. It is a well-known phenomenon that anxiety and depression are frequently underreported and under-diagnosed.^22,23 One possible explanation is that the somatic symptoms are most often the focus for treatment. Hence, this is the primary concern of both physicians and patients.^22,23 In addition, some patients do not expect to receive treatment for their emotional problems, as these may be regarded as an inevitable or natural consequences of having advanced cancer.^3,23,24 It may also be that the patients regard depression or anxiety as psychiatric disorders, thereby implying a tendency toward scoring lower on these scales. This corresponds with findings both in the study of Watanabe et al. ¹⁷ and in the present study, as some of the patients found it important to express that they were not depressed in the diagnostic sense of the word. In the Norwegian version of the ESAS more everyday terms have been added to the original anchors: anxiety/uneasiness and depression/sadness. This seemed to lead to another problem; the dual meaning of the questions made it difficult for some patients to decide on how to respond. Survey research recommends avoiding such equivocal questions because they create confusion and makes it difficult to interpret the answers, thereby reducing the validity. ¹⁶

Previous work has documented that patients find it particularly difficult to understand the question of well-being.^11,17,25 Watanabe et al. ¹⁷ suggested placing this specific question at the end of the tool, because it comprises all the other symptoms. This is supported by the results of our study. The placement of ‘well-being’ at the end of the tool seemed to have an encapsulating effect. Many patients related their response on this specific question to their previous answers. This may explain the observed differences between our results and other findings. An alternative explanation could be due to the Norwegian translation of this concept. The exact term ‘well-being’ is rarely used in Norwegian and is translated into ‘Overall, how are you feeling today?’, which is a more unequivocal question. None of the Norwegian participants found it difficult to answer this question.

Comprehension is not only influenced by the exact wording of a symptom. The response format also has an effect. This became apparent with the appetite item, for which some patients interpreted low scores as equivalent to little appetite, without realizing that their answer actually reflected the opposite. This finding corresponds with previous research on the ESAS^12,17 that has suggested changing the terminology and the format of some of the items in order to obtain consistent scale scores throughout the form. For this particular item, the scale is reversed, with zero indicating normal appetite or no appetite loss. Research has shown that the number zero tends to be given a logical meaning, such as the absence of a phenomenon, when used within a response scale. ¹⁶ What may happen is that some patients interpret that the numeric value of 0 indicates absence and as such overrides (makes them overlook) the anchoring term ‘good’ connected to the symptom. Furthermore, health care providers have been advised to be aware of the problem and hence they should review the scores together with the patients ¹² or be present when the ESAS is being completed. ¹⁷ The pros and cons of health staff being present while patients fill in the ESAS have been discussed in previous studies.^12,17 When being present during the completion process, it is important that the health care provider is aware of the problem of social desirability and consider this up against the positive effects of avoiding misinterpretations.

Studies on pain assessment have pointed out that the anchoring term ‘worst possible’ has only rarely been elaborated on from the patients’ perspective. ²⁶ Some patients have found it difficult to rate worst possible pain, because imagining worst possible pain has activated too many negative thoughts and feelings. ²⁷ In our study this aspect was not just related to pain, but also to the psychological symptoms (anxiety and depression). This may be due to the complexity of these symptoms. It could also be difficult to imagine worst possible because this suggests a future perspective, invoking the possibility that things can always become worse. These arguments support the need for health staff to frequently check whether the patients comprehend ‘worst possible’ as intended.

Our results also showed individual differences in how the numerical values were interpreted. The same numerical value could represent different intensity of a symptom, and similar symptom descriptions were represented by different numerical values. Similar findings were also described in a recent study, ²⁸ where it was examined what patients meant when assigning words or numbers to the symptoms of the ESAS. For example, with the symptom of pain they found that the word ‘mild’ was given numerical responses from 0 to 8 and the category ‘moderate’ ranged all the way from 0 to 10. Although it is not the purpose of the ESAS to compare ratings between patients, such variability in ratings could cause misunderstandings among the health care providers if the individual patient’s interpretation is not taken into consideration. If the health staff base their evaluation of the ESAS scores on either their own thoughts about severity or on threshold values, this could affect further treatment and result in inexpedient symptom management.^12,17,28 On the other hand, the patient is the most important source of information regarding subjective symptoms and with repeated measures over time changes in the individual symptom burden can be monitored.

As with most subjective measures, some lack of consistency in the individual patients’ scores also occurred. During the response process, previous symptom experience and experience with the ESAS tool as such, as well as the actual context, will affect the rating of the symptoms. The interviews revealed that factors such as mood, specific events and time of the day could influence the response at a specific point of time. This indicates that different scores over time may not only be a result of an objective variation in symptom intensity or an effect of treatment, but could also be influenced by other aspects. Particularly with regard to pain, research has shown that pain intensity and its often fluctuating nature represent a complex phenomenon that is often affected by the context.^29,30 Our results showed that this was also the case for the two psychological symptoms, anxiety and depression, as the patients found it difficult to comprehend and rate these symptoms without relating their answers to the context. This was also the case in the study of Watanabe et al., ¹⁷ where some patients commented that how they scored anxiety and depression could often change depending on the context.

A shortcoming of our study is that the Norwegian modifications to the ESAS may limit the comparison with other studies. Another possible limitation is that maybe only the healthiest patients agreed to participate. Thus it could be that the difficulties with the ESAS could have been more pronounced among patients with a higher symptom burden. Whether the use of only an inpatient sample is a limitation or not, is difficult to assess. Furthermore, we do not know if the patients who volunteered for the study did so because they were particularly satisfied or dissatisfied with the tool, which in turn could reduce the transferability of the results in either way. In a recent study it was found that men and women did respond to the ESAS in different ways. ³¹ Thus it could be that the disproportional gender balance (W:3/M:8) in our study may have biased the results. Many of the patients were tired during the interview and were sometimes unable to elaborate on the different topics. In addition, the quality of the interviews will always depend on the informants’ verbal ability and the relationship between the interviewer and the informant. ¹⁶

The widespread use of the ESAS, first and foremost as a symptom screening tool, is mainly because of its brevity and applicability in most settings. Our results indicate that the validity and clinical utility of ESAS would be further improved by reducing the risks of misunderstanding. This could be achieved by implementing the following actions: (1) make sure that the patients are familiar with the purpose of the tool; (2) sit down and carefully go through the completed form together with the patient; and (3) provide feedback on the patients’ answers. In our opinion, a stronger focus on how to standardize the use of the ESAS may contribute to improve its usefulness in palliative care.

Conclusion

The present study has shown that errors and misunderstandings do occur while completing the ESAS. This underlines the importance of examining the patients’ interpretations. A standardized method for using the ESAS is a prerequisite to reduce the risk of errors, which may increase the clinical utility of the tool and improve symptom management.