Breast Cancer Research and Treatment

, Volume 134, Issue 3, pp 1327–1335

The impact of an electronic health questionnaire on symptom management and behavior reporting for breast cancer survivors

Authors

  • Meredith Bock
    • School of MedicineUniversity of California, San Francisco
  • Dan Moore
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • Jimmy Hwang
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • Dianne Shumay
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • Laurell Lawson
    • Dynamic Clinical Systems
  • Deborah Hamolsky
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • Laura Esserman
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • Hope Rugo
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • A. Jo Chien
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • John Park
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
  • Pamela Munster
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
    • Helen Diller Family Comprehensive Cancer CenterUniversity of California, San Francisco
    • Department of Medicine (Hematology/Oncology)University of California, San Francisco
Epidemiology

DOI: 10.1007/s10549-012-2150-1

Cite this article as:
Bock, M., Moore, D., Hwang, J. et al. Breast Cancer Res Treat (2012) 134: 1327. doi:10.1007/s10549-012-2150-1

Abstract

Breast cancer (BC) patients experience multiple symptoms as a result of diagnosis and treatment. While surveillance for detecting cancer recurrence is fundamental to follow-up care, managing symptoms, and promoting health behaviors are equally important. UCSF has implemented a secure online health questionnaire enabling BC patients to provide updates of their health history and symptoms. We randomly selected a sample of stage I–III BC patients (n = 106) who completed a questionnaire before a medical oncology visit between August 2010 and January 2011 and consented to have data used for research. We conducted a chart review calculating the number of symptoms reported in the questionnaire, the clinic note only, and both questionnaire and clinic note, excluding chronic symptoms addressed previously. Self-reported data on exercise and alcohol consumption was compared to documentation of these lifestyle factors in clinic notes. Patients reported significantly more symptoms using the online questionnaire (mean = 3.8, range 0–13) than were documented by the provider in clinic notes (mean = 1.8, range 0–7; p < 0.001 for the difference). A regression plot comparing the percentage of symptoms agreed upon by the patient and provider and the percentage of symptoms addressed yields a slope of 0.56 (95 % CI 0.41–0.71). The number of self-reported symptoms correlates with self-reported Karnofsky scale such that the number of symptoms reported by the patient increases linearly with this score until a threshold and it then plateaus (p < 0.001). Exercise behavior and alcohol consumption were reported in 100 % of the online questionnaires, but was documented in only 30/106 (28 %) and 75/106 (70 %) of charts reviewed. In 19/75 (25 %) charts with alcohol consumption documented, there was substantial discordance between patient and clinician reporting. Electronic data collection of BC patient-reported outcomes has a positive effect on symptom management and identification of opportunities for risk-reducing behavior change.

Keywords

Patient-reported outcomesBreast cancerClinical practiceHealth information technology

Introduction

Breast cancer patients experience a complex array of symptoms from their disease and its treatment. While detection of cancer recurrence is fundamental to follow-up care, effective symptom management and the discussion of risk-reducing lifestyle change are equally important. The routine collection and assessment of patient-reported outcomes (PROs) is a promising method to facilitate symptom management and behavioral interventions.

The incorporation of patient-reported symptoms and quality of life measures into clinical practice has been validated as a viable tool for improving patient-provider communication. Electronic data collection systems are feasible and associated with high patient satisfaction [13]. Patients and clinicians have expressed that PROs are useful in identifying under-reported issues, focusing the clinic visit, and tracking symptoms over time [4].

Based on these established benefits, investigators have called for the standardization of collecting PROs [57]. However, though studies have indicated that routine electronic assessments lead to an increased discussion of quality of life issues [810], the results to date are less definitive as to how these systems affect patient management [11]. The primary purpose of this study was to investigate the impact of a web-based health questionnaire on symptom reporting, physician documentation of symptoms, and symptom management. The secondary aim was to compare self-reported data on exercise and alcohol consumption to the clinician documentation of these lifestyle factors.

Methods

The UCSF Carol Buck Breast Care Center implemented a secure web-based health questionnaire enabling patients to provide and update their health history and symptoms upon establishing care at our center and before each follow-up clinic visit. Integrated Survey System® (ISS) Software by Dynamic Clinical Systems is used for data collection and outcomes reporting. Each appointment in the UCSF scheduling system generates an automated email request inviting the patient to complete the online survey. A comprehensive initial survey is requested before each new patient appointment and a shorter follow-up request is generated for subsequent appointments. Extensive branching logic insures that an appropriate version of the questionnaire can be extended to all patients whether they are receiving preventative care or active treatment.

If the patient does not have internet access, portable tablet computers are available in the clinic waiting room before their appointment. Embedded within the web-based health questionnaire, patients are presented with an electronic consent document asking permission to have their health history and symptom data used either in a de-identified fashion or linked to their clinic chart and other medical records for research purposes. A report summarizing this patient-reported information is placed in the chart for clinician review before visits. Figure 1 depicts the process by which PRO data are collected and used clinically.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2150-1/MediaObjects/10549_2012_2150_Fig1_HTML.gif
Fig. 1

Automated web-based questionnaire process at the UCSF Breast Care Center

The clinician summary report placed within the chart before the clinic visit contains the patient’s self-reported treatment history, demographic information, symptoms, and health behaviors. Symptoms are listed in a standard “Review of Systems” (ROS) format. If a patient advocates having a particular symptom, the patient is then queried for the frequency, severity, and distress associated with the symptom. Frequency is rated by the patient as “rarely,” “occasionally,” “frequently,” or “almost constantly.” Severity is rated as “slight,” “moderate,” “severe,” or “very severe.” Distress is rated as “not at all,” “a little bit,” “somewhat,” “quite a bit,” or “very much.” A symptom is additionally placed in a “Red Flag” (RF) category at the top of the survey if it had not been reported previously or its associated frequency, severity, or distress increased by at least two degrees since the last visit.

For this study, we selected a sample of stage I–III breast cancer patients (n = 106) who completed the follow-up questionnaire before a medical oncology visit occurring over a period of ~6 months between August 2010 and January 2011. The compliance for survey completion rate is nearly 80 % for new patients and ~40 % for follow-up patients. Approximately 735 follow-up patients completed the survey and consented to have their identified data used for research during this time period. The sample of 106 patients was selected from this pool. The investigators observed that the sample fairly represented the clinic volumes of the medical oncology clinic providers including physicians (n = 7) and nurse practitioners (n = 3). The sample is also representative of the distribution of treatments (measured by hormone therapy only, chemotherapy only, or both) for stage I–III patients in the clinic. In these respects, this sample is a reasonable representation of all English-speaking, non-metastatic patients in follow-up care in our clinic.

Two of the investigators (MB and MM) conducted a chart review calculating the number of symptoms reported by the patient in the questionnaire, documented in the clinic note by the care provider (either a physician or nurse practitioner), or both reported in the questionnaire and documented in the clinic note. We also assessed whether each symptom was addressed during the appointment. A symptom was considered managed if an intervention, prescription, or referral was documented in the clinic note. During this chart review, we excluded chronic symptoms addressed previously (within a year of the appointment under review or during two prior clinic visits, whichever constituted a longer period). Patients’ treatment status and self-reported Karnofsky scores were also collected to investigate their relationship with symptom reporting and management.

We used a paired t test to compare the number of symptoms reported in the questionnaire compared to those documented in the clinic note by the clinician. A regression plot was used to determine whether the fraction of symptoms agreed upon by the patient questionnaire and provider clinic note correlated with symptom management. Self-reported symptoms are summarized by whether they were documented in the clinic note, managed, or both documented in the note and managed according to treatment history or Karnofsky score. Symptoms reported on the survey were also subdivided into ROS and RF categories to determine if severity affected likelihood of symptom management.

Comparisons of symptom documentation and management between groups of interest are performed using Fisher’s exact test. The linear trend of the symptom documentation and management with Karnofsky score is evaluated using an F test.

In addition, self-reported data on exercise and alcohol consumption was compared to documentation of these lifestyle factors in clinic notes. A comparison was considered “discordant” if the amount of exercise or alcohol consumption indicated by the patient differed substantially from the provider-documented amount. For example, if a patient documented drinking one to two alcoholic beverages three to four times per week, but the provider documented “none” or “rare” alcohol consumption, this was considered discordant. Alcohol consumption and exercise levels were chosen for analysis because they are modifiable behaviors that have been associated with breast cancer recurrence. Frequency of smoking is also documented on the survey, but the analysis of this variable was not expected to be informative as the percentage of smokers within this clinic is low.

Results

Characteristics of the patients in the sample are shown in Table 1. Table 2 enumerates the most commonly reported symptoms, which include joint pain or stiffness (n = 45), hot flashes (n = 39), fatigue or lack of energy (n = 35), anxiety (n = 33), difficulty with concentration or memory (n = 31), and decreased sexual interest or vaginal dryness (n = 28). Of the six most commonly electronically reported symptoms, joint pain, fatigue, anxiety, and vaginal dryness were managed over 20 % of the time. However, hot flashes and difficulty concentrating/memory loss had a much lower probability of being managed (5 and 10 %, respectively). Hot flashes account for 21 % of all electronically reported symptoms that were also documented but not managed by the provider. Twenty-seven out of the 31 instances (87 %) of patient-reported difficulty concentrating were neither documented in the clinic note nor managed. Whereas hot flashes were frequently documented but not managed, providers typically did not document or manage patient-reported difficulty concentrating.
Table 1

Demographics of patient sample

Characteristic

No.

 

Percentage

Age, years

 Mean

 

56.9

 

 Range

 

32–85

 

Race

 White

92

 

87

 Black

3

 

3

 Chinese

3

 

3

 Japanese

1

 

1

 Other

7

 

6

Ethnicity

 Non-Hispanic

97

 

91

 Hispanic

2

 

2

 Not reported

7

 

7

Table 2

Mode of report and management of common symptoms (n > 5)

Symptom

Reported in survey only

Reported in survey and clinic note

Total

Managed

Not managed

Managed

Not managed

No.

%

No.

%

No.

%

No.

%

Joint pain or stiffness

1

12.5

22

5.9

13

11.6

9

8.2

45

Hot flashes

0

0

14

3.8

2

1.8

23

20.9

39

Fatigue, lack of energy

0

0

18

4.9

10

8.9

7

6.4

35

Anxiety

1

12.5

18

4.9

8

7.1

6

5.5

33

Difficulty with concentration/memory

0

0

27

7.3

3

2.7

1

0.9

31

Decreased sexual interest/vaginal dryness

2

25

18

4.9

6

5.4

2

1.8

28

Difficulty sleeping

1

12.5

11

3

8

7.1

6

5.5

26

Dry skin or itching

0

0

18

4.9

1

0.9

4

3.6

23

Hair loss

0

0

16

4.3

2

1.8

0

0

18

Dry eyes

0

0

11

3

2

1.8

3

2.7

16

Night sweats

0

0

13

3.5

1

0.9

2

1.8

16

Frequent urination/incontinence

1

12.5

15

4.1

1

0.9

2

1.8

15

Numbness or tingling

0

0

9

2.4

2

1.8

4

3.6

15

Back Pain

0

0

7

1.9

3

2.7

4

3.6

14

Cough

0

0

12

3.2

2

1.8

0

0

14

Depression

1

12.5

6

1.6

5

4.5

2

1.8

14

Constipation

0

0

10

2.7

1

0.9

0

0

12

Difficulty swallowing

0

0

8

2.2

3

2.7

1

0.9

12

Feeling bloated

0

0

10

2.7

1

0.9

0

0

11

Shortness of breath or wheezing

0

0

5

1.4

4

3.6

1

0.9

10

Chest pain, tightness

0

0

6

1.6

3

2.7

0

0

9

Pain

1

12.5

3

0.8

3

2.7

2

1.8

9

Blurry or double vision

0

0

4

1.1

2

1.8

2

1.8

8

Bruises or bleeds easily

0

0

6

1.6

0

0

1

0.9

7

Hearing difficulty

0

0

2

0.5

2

1.8

1

0.9

7

Heartburn, indigestion

0

0

4

1.1

1

0.9

2

1.8

7

Sweats

0

0

3

0.8

1

0.9

3

2.7

7

Dry mouth

0

0

6

1.6

0

0

0

0

6

Swelling of arms or legs

0

0

5

1.4

1

0.9

0

0

6

Total

8

100

307

83.1

91

81.5

88

79.8

493

Note: The “%” column reflects the frequency of each symptom in the corresponding category of documentation and management. Not all total percentages equal 100 % because the denominator was defined as all non-chronic symptoms reported (n = 600) in the patient sample. Only the most common (n = 493) are reported in this chart

Symptom management

Symptom management data are presented in Fig. 2. Patients reported significantly more symptoms using the online questionnaire (mean = 3.8, range = 0–13) than were documented by the provider in clinic notes (mean = 1.8, range 0–7; p < 0.001 for the difference according to a paired t test). A regression plot (Fig. 3) summarizes the relationship between the percentage of symptoms agreed upon by the patient and provider and the percentage of symptoms addressed, yielding a slope of 0.56 (95 % CI 0.41–0.71). For each 1 % increase in symptom agreement between patient and provider, there is a 0.56 % increase in the percentage of symptoms that are addressed during the consultation. A slope of 0.56 indicates that about half the symptoms mentioned by both the clinician and the patient are addressed, regardless of the number of symptoms.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2150-1/MediaObjects/10549_2012_2150_Fig2_HTML.gif
Fig. 2

Number of symptoms by mode of report

https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2150-1/MediaObjects/10549_2012_2150_Fig3_HTML.gif
Fig. 3

Lowess plot of the number of symptoms addressed versus the number reported by both the patient questionnaire and the clinician notes. Each dot indicates data from one patient. Dots are jittered to show individual data points. Regression shows that the linear term is significant (p = 0.015) while a quadratic term is not (p = 0.22). The adjusted R2 is 0.50 indicating that the regression line accounts for 50 % of the variation

The number of self-reported symptoms increases with perceived functional impairment as measured by the self-reported Karnofsky scale. The locally weighted least-squares plot of number of symptoms versus Karnofsky score (Fig. 4) suggests that the number of symptoms reported by the patient increases with score up to a score of 3 and then levels off. The linear increase with scores 1–3 is significant (p < 0.001 by linear regression analysis). However, lower percentages of the symptoms reported by patients with higher Karnofsky scores are documented by the provider or managed (Table 3). The percentage of electronically reported symptoms that are both documented in the clinic note and managed to decrease with rising Karnofsky score in a linear fashion (r = 0.838, p = 0.016).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2150-1/MediaObjects/10549_2012_2150_Fig4_HTML.gif
Fig. 4

Locally weighted least-squares fit (lowess) to number of symptoms reported by the patient versus self-reported Karnofsky score (p < 0.001 for the linearly increasing portion with score 1–3)

Table 3

Symptom documentation in clinic note and management by patient characteristic

 

No. patients

Documented

Managed

Documented and managed

Mean %

SD

Mean %

SD

Mean %

SD

Treatment

 None

3

41

53

26

36

39

39

 Hormone therapy

39

36

32

19

25

36

25

 Chemotherapy

14

37

29

17

17

24

10

 Hor. and Chemotherapy

50

37

32

22

29

36

25

Karnofsky score

 1

40

33

40

15

28

45

32

 2

47

38

28

21

24

33

19

 3

11

40

27

35

30

32

31

 4–6

6

34

24

20

18

29

15

Treatment history does not appear to influence the likelihood that a self-reported symptom is documented in the clinic note or whether a symptom is managed (Table 3). However, there is a trend that a significantly higher proportion of symptoms are both documented in the clinic note and managed if the treatment history includes hormone therapy. An average of 24 % of symptoms were managed for patients who were treated with chemotherapy alone compared with 36 % of symptoms reported by patients treated with hormone therapy alone or both chemotherapy and hormone therapy (p = 0.085). This trend becomes more significant when symptoms are further subdivided based on severity (Table 4). For all symptoms, patients who were treated with hormone therapy had a consistently higher percentage of their symptoms managed. For RF symptoms, patients treated with hormone therapy had an average of 28 % of their symptoms managed. This was significantly higher than the percentage of symptoms managed for patients treated with chemotherapy alone (18 %) or both chemotherapy and hormone therapy (19 %, p = 0.007 for the comparison). Table 4 also displays that symptoms red-flagged on the survey were not more likely to be documented in the note and/or managed by the provider.
Table 4

Symptom documentation in clinic note and management by severity

Treatment

No. patients

ROS (less severe)

RF (more severe)

Documented, not managed

Documented, managed

Documented, not managed

Documented, managed

%

SD

%

SD

%

SD

%

SD

Hormone therapy

39

21

13

32

29

20

10

28

20

Chemotherapy

14

32

17

17

14

13

3

18

10

Hor. and Chemo.

50

25

13

30

29

21

14

19

10

Total

103

24

13

29

27

19

11

22

14

Note: The patients that received neither hormone therapy nor chemotherapy (n = 3) were excluded from this analysis

Health behaviors

Information on exercise behavior and alcohol consumption was completed by the patient in 100 % of the online questionnaires, but was documented by the clinician in only 30/106 (28 %) and 75/106 (70 %), respectively, of charts reviewed (see Figs. 5 and 6). In 19/75 (25 %) charts where alcohol consumption was documented, there was substantial discordance between patient and clinician reporting. In all 19 cases, the provider documented a lower amount than what the patient reported in the survey.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2150-1/MediaObjects/10549_2012_2150_Fig5_HTML.gif
Fig. 5

Frequency of alcohol consumption documentation by physician in clinic note compared to self-reported levels

https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2150-1/MediaObjects/10549_2012_2150_Fig6_HTML.gif
Fig. 6

Frequency of exercise behaviors documentation by the physician compared to self-reported levels

Discussion

The results from this investigation suggest that the addition of an online health questionnaire into clinical practice elicits an increase in symptom reporting. We also discovered that roughly half the symptoms reported by both the online questionnaire and the clinical notes are addressed, regardless of total number of symptoms reported. Online health questionnaires appear to facilitate patient-provider communication by providing the clinician a more comprehensive list of symptoms being experienced by the patient within a short time frame before the clinic encounter. This finding is similar to the results based on 43 breast cancer patients from a randomized controlled trial by Berry et al. [10]. Their study also found that use of a patient self-assessment questionnaire improved treatment of symptoms without lengthening the clinic visit.

Our study noted that patients treated with hormone therapy alone had a higher proportion of their symptoms managed than those who were treated with chemotherapy. This difference was most pronounced for symptoms that were new or particularly distressing for the patient. This may reflect the availability of effective treatments for the most common symptoms associated with hormone therapy. In support of this, several of the most common symptoms reported in this study are often related to hormone therapy and—with the exception of hot flashes—show relatively higher rates of documentation and management. This difference may also be explained by a shift in the provider’s perspective to focus on symptom management in a cohort with a lower risk of recurrence.

Our study also noted a correlation between the number of symptoms reported by the patients and the self-reported Karnofsky score, a separate measure of functional impairment. We observed that the number of symptoms reported electronically by patients’ increases up to a certain threshold and then plateaus as the self-reported Karnofsy score continues to worsen. This plateau effect may be a result of the exclusion from this analysis of chronic symptoms related to other co-morbid conditions or previously managed symptoms. Persistent and potentially debilitating symptoms that are longstanding or related to co-morbidities other than breast cancer were not included in our analysis. In addition, chronic symptoms relating to breast cancer treatment such as neuropathy and hot flashes were not counted if they had been addressed in one of two previous clinic visits or within the past year. Management of these symptoms is often not optimal and they continue to impact patients’ quality of life and contribute to functional impairment. This plateau illustrates that the number of symptoms reported by a patient may not be as important as the distress caused by individual symptoms. Nonetheless, this study suggests that because the number of self-reported symptoms on the survey correlates with the self-reported Karnofsky, this score appears to be a useful single item measure to understand how symptomatic patients perceive themselves to be and could be incorporated into clinical care more routinely to make a quick assessment of patients’ functional well-being.

Interestingly, an additional linear correlation was found between self-reported Karnofsky score and likelihood of symptom documentation and management. Patients that self-identified as more functionally impaired had a consistently lower proportion of their symptoms managed. In a related finding, symptoms experienced as new or particularly distressing to the patient are less likely to be managed by the provider. This may reflect the increased complexity and lack of sufficient time for symptom management in patients with multiple co-morbidities or other sources of functional impairment. The failure to manage new or particularly distressing symptoms may also be explained by the provider’s focus on determining symptom cause to rule out cancer recurrence.

Health behaviors were more consistently reported in the online health questionnaire compared to the clinic note. Based on this chart review, lifestyle factors may be under-reported. Online health questionnaires can increase the consistency of screening for at-risk behaviors and may enable providers to address relevant behavior change. Obesity and weight gain have been well-described as risk factors for developing breast cancer and as risk factors for recurrence after diagnosis and treatment of breast cancer [1216]. Exercise has also been associated with reduction in risk of breast cancer recurrence [17, 18]. Physical activity also plays a central role in weight maintenance that contributes to the reduction of overall mortality [19]. Lack of documentation of exercise behavior is a missed opportunity in providing comprehensive health care in a cancer survivor. Similarly, excess alcohol consumption has been shown to be associated with risk of developing breast cancer and also increased risk of recurrence among certain subsets of breast cancer patients [20, 21]. Based on the clinic notes alone, alcohol consumption was found to be missing or discordant for 1 in 3 clinic visits in the Breast Care Center. Interestingly, in the discordant cases the clinician documented less than what the patient reported in the survey, suggesting that patients report lower amounts of alcohol consumption in a face-to-face encounter with their provider. Each clinic visit provides an opportunity for the clinician to suggest lifestyle modification and our study suggests that opportunity is often missed due to lack of information. In the primary care setting, a brief counseling intervention for high-risk drinkers was found to be effective in reducing alcohol intake [22]. Given that many cancer patients are motivated to make lifestyle changes, particularly if there is some evidence that it may impact their risk of recurrence, it is likely that counseling on alcohol intake would be successful in the breast cancer population as well.

The collection of PROs has long been considered an important step in increasing the clinical emphasis on health-related quality of life [2325]. Prospective randomized trials have demonstrated that the use of a graphical summary of patient-reported information improved discussion of quality of life issues This study complements the current literature by demonstrating that PROs, in addition to the increasing discussion of problematic symptoms, may actually lead to an increase in the number of symptoms addressed and managed. It has been hypothesized in the past that electronic systems provide a “safe environment,” where patients can more accurately report on sensitive topics [26]. Our finding on health behaviors bolsters this claim and provides evidence of disparities between patient-reported and clinician-documented levels of alcohol consumption and exercise, two behaviors with well-described effects on breast cancer incidence and recurrence. The electronic survey also provides patients with more time to report a comprehensive list of symptoms and behaviors, enabling the provider to prioritize during the limited time of the consultation. Though much of the current literature about these web-based tools focuses on how they promote discussion of symptoms in later stage cancers [27, 28], this analysis also supports their effectiveness in managing patients with curable disease.

This retrospective analysis has a number of limitations. Our study was not a prospective and patients were not randomized to treatment groups. We did not audio-record the clinic visits; thus we cannot compare whether the number of symptoms reported electronically equals or exceeds the number of symptoms that a patient typically reports to their care provider verbally. Since the electronic surveys are completed from days to several weeks in advance of the clinic visit, we also cannot ascertain whether a symptom reported by the patient had resolved by the time of the visit. Data regarding whether a symptom was addressed and managed was limited to the clinician documentation in the clinic note, and it is possible that the provider reviewed the electronically reported symptoms and only chose to document the symptoms that the patient identified in the appointment as the most severe. It is also possible that the patient was provided with a prescription for symptom management or a test was ordered to evaluate the symptom without adequate documentation in the note. Another possible contributor to the variability in symptom management is the medical oncologists’ attitude toward managing symptoms unrelated to cancer, and it is possible that the oncology care providers in this study did not document symptoms that they felt should be managed by the primary care physician. In some cases, patients rely on their medical oncologist to act as their primary care physician, whereas other patients with multiple medical problems may rely more heavily on their primary care physician or other specialists to manage symptoms not directed related to cancer. Finally, since the questionnaires are not available in multiple languages, only patients who are English speaking or have an English-speaking surrogate to complete the survey are included in this sample. Therefore, our findings may not apply to non-English-speaking patients.

This retrospective study cannot account for the substantial variation in symptom management among different providers and for different patients even in the same practice setting. Despite optimization of patient-provider communication regarding quality of life issues, some symptoms will remain incompletely managed due to lack of successful interventions available to treat certain cancer and treatment-related problems. However, this study does illustrate a positive effect of electronic data collection of PROs on symptom reporting. While not the aim our study, compiling data from patient electronic questionnaires on the most frequently reported, problematic and distressing symptoms certainly illustrates the need for additional research, funding, and resources to improve symptom management in these areas. Our study also demonstrates the utility of PRO in identifying at-risk behaviors such as excess alcohol consumption and sedentary lifestyle and provides the opportunity to intervene and facilitate behavior change. Further research is necessary to determine if increasing symptom reporting through PRO can actually improve provider documentation, patient quality of life, and overall health outcomes.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer Science+Business Media, LLC. 2012