The expanded prostate cancer index composite short form (EPIC-26) for measuring health-related quality of life: content analysis of patients’ spontaneous comments written in survey margins

Measurement instruments are necessary tools for research and clinical practice [1]. Choosing a quantitative instrument to assess patient-reported outcomes, however, can be challenging, since all research instruments have limitations. Measuring quality of life (QoL) is particularly challenging. QoL is subjective, multidimensional, and complex, and even the best instruments can capture only a slice of its entirety [2, 3]. Health-related QoL (HRQoL) includes physical, psychological, and social dimensions that impact one’s health [4]. Assessment of HRQoL is particularly important in healthcare, since many diseases and treatments can affect patients’ perceived health and well-being. This is especially important regarding diseases where treatment decisions may be affected by expected QoL outcomes [3]. Therefore, it is essential that QoL instruments are valid, reliable, and psychometrically sound [4].

Several validated and extensively used QoL instruments have proved suitable for HRQoL evaluation with prostate cancer patients. The most frequently used HRQoL instruments are the 50-item Expanded Prostate Cancer Index Composite (EPIC-50) [5], its short form (EPIC-26) [6, 7], and the 16-item EPIC for Clinical Practice (EPIC-CP) [8] for routine clinical use. These instruments include five symptom domains—urinary incontinence, urinary irritation, bowel, sexual, and hormonal—that relate to common prostate cancer treatment side effects and related bother. EPIC-26, which was also used in this study, includes 26 items: three items for urinary function and six for urinary bother, six items for bowel bother, five items for sexual function and one for sexual bother, and five items for hormonal bother. No validity or reliability issues have been found within this instrument [7]. Although EPIC-50 was developed in cooperation with patient experts [5], documentation of users’ experience with these EPIC-50-based instruments is scarce.

Korzeniowski et al. [9] interviewed clinicians about their experiences of using EPIC-26 in practice, perceptions of the instrument’s value, operational issues, and impressions of patient acceptability and value. These clinicians’ patients’ comments on ease of use and sources of confusion were also recorded. The authors found that patients were mostly willing to report their disease-specific symptoms to their healthcare team, since they felt this would affect their cancer care positively. The study revealed that a few patients with pretreatment sexual limitations faced barriers when answering questions related to sexual functioning, as a “not applicable” category was missing.

Because, to our knowledge, Korzeniowski et. al. are the only researchers to address user experience of EPIC-26 so far, we checked what there is to learn about other EPIC-50-based measures. Brundage et al. [10] evaluated the implementation of EPIC-CP in clinical practice for men with early-stage prostate cancer by exploring the usability and acceptability of the questionnaire through a patient “exit survey.” They also interviewed clinicians about EPIC-CP use. This multimethod study confirmed that the “majority of patients rated use of EPIC-CP favorably and also clinicians’ attitudes were supportive.” However, information about pragmatic usability was limited to patients’ survey answers regarding touchscreen or software use, font size, printing the report, images, place of answering the survey, and help received to answer the survey. A need for further information about why patients skip questions was recognized, although item completion rates were generally high (91–100%). In relation to our analysis, it is notable that the lowest completion rates in their study were for questions concerning sexual health.

We chose to use EPIC-26 in our research project evaluating long-term QoL changes among prostate cancer patients because it was already highly used in prostate cancer care in Finland. To our surprise, 22% of survey participants wrote comments in the margins of the EPIC questionnaire. The large number of remarks beside EPIC questions led us to consider whether men might have difficulties finding suitable answer options, or whether the questionnaire neglects matters that patients consider significant. Unlike studies concerning physicians’ views of EPIC, no previously published research uses qualitative methods to explore patients’ perceptions of EPIC instruments. Considering respondents’ feedback is crucial for the development of any instrument’s reliability and validity. Therefore, in this article, we aim to depict patients’ perspectives on the use of EPIC-26 (see Online Appendix).

This study investigates the comments that prostate cancer patients write in the margins of the EPIC-26 questionnaire. Our goal is to use spontaneously generated qualitative data to show the possible barriers that patients face while answering EPIC-26, and to consider how these barriers may affect the response data generated from the instrument. Based on our findings, we also suggest some future directions to improve the quality of the instrument and minimize the number of skipped questions.

Our research questions are as follows:

This study aimed to depict patients’ perspectives on the use of EPIC-26 to measure HRQoL after prostate cancer diagnosis. In respect to further enhancing the reliability and validity of this widely used instrument, we were interested in respondents’ experiences of the survey. Here, we discuss our key observations and provide suggestions for future directions in the development of EPIC-26.

Brundage et al. [10] recognized the need to determine strategies to minimize missing data in EPIC-CP, and wished for information regarding why patients skip questions. They also noted that missing data were especially a problem in the survey’s sexual domain. Similarly, our participants gave several reasons for skipping questions, and the sexual domain in particular produced more than half of all comments. A major reason for skipping items was that the question was not relevant to the patient if he had had no sexual activity during the previous four weeks. The problems with answering erectile function questions lie in the questions’ presumptions. Question 10 assumes that the respondent wants to have an erection (see Appendix), but this is not always the case. It is also important to acknowledge that if one has not been sexually active in the previous four weeks, it is difficult to estimate one’s erectile ability (questions 8–9). Our study supports previous findings that patients require a “not applicable” category for erectile function questions [9]. The same issue does not pertain to the sexual bother question, since it does not presuppose sexual activity or desire.

Erectile medication use is not considered in this disease-specific instrument. In addition to the difficulty it causes patients, who have to decide whether or not they are expected to take their medicine use into account, this shortcoming affects the interpretation and reliability of the statistical data received. First, any difficulty in completing a survey increases the risk of missing data. Second, if the patient does not recall whether he took his erectile aid use into account during earlier survey rounds, he may use different strategies to answer the same question in different rounds; this will weaken the test–retest reliability, although EPIC-26’s reliability has been confirmed as sufficient [7]. Hence, the use of erectile aid should be incorporated into this instrument targeted at prostate cancer patients.

In the urinary domain, a clear reason for skipping questions was the use of an indwelling or suprapubic urinary catheter, which made the questions irrelevant to the patient. Due to prostate cancer treatment or progression of the cancer, catheter use is often required. Hence, it would be reasonable to note catheter use in the questionnaire. Although the same issue emerged with colostomies, the number of these patients is smaller, and it might not be so reasonable to take this matter into account and thereby increase the length of the survey.

Because quantitative surveys cannot consider all the possible factors that patients might consider important [2], there are always going to be comments written in the margins of pen-and-paper surveys. Patients’ descriptions of their health condition generated the largest theme in our study, illustrating patients’ natural need to explain and clarify their life experiences. Compared with electronic surveys with no space for additional information, the ability to write in a paper margin might encourage respondents to answer a survey, since there is space for explanation with every question. That said, a computerized version of EPIC-26 has also proved to be highly accepted by patients in clinical practice [13]. As we consider these e-versions useful tools, free-text fields are important in online surveys where there is otherwise no natural space for comments. Allowing comments also allows transpiring of previously unknown problems within the survey and participants development suggestions that are valuable for the development of the measure.

This study sheds light on the challenges related to measuring HRQoL with a survey that focuses on disease-related symptoms and bother. Our study concerns mainly physical dimension of patients’ overall QoL with the main findings directing to sexual and urinary functioning. Our findings regarding EPIC-26 are in line with the previous results on EPIC-CP, in that patients are mainly willing to report their symptoms [10]. The challenges related to answering EPIC-26 relate to a lack of suitable answer options, which might weaken this instrument’s content validity if the items are insufficiently comprehensive or relevant for certain patients. Our conclusion is that considering the use of erectile medication and adding a new category to sexual function questions 8–10 would decrease the skipping of sexuality-related questions and further improve the quality of EPIC-26. Similarly, as mentioned earlier, a consideration of the use of urinary catheters would probably reduce missing data on items 1–4. Certainly, further studies are required to examine the statistical aspects of these suggested developments. Also, broader statistical analyses on sociodemographic or clinical factors related to skipping questions in EPIC would be important to gain more knowledge about which patient groups struggle the most in finding suitable answer options.