patient recall accuracy

Cancer Patient Recall of Doctor Communications

The bulk of research on doctor-patient communications have suggested a positive association between effective communication and patient health outcomes. Yet how much can the patients recall about discussions with their doctors, and how much does that memory recede over time? 

This analysis of close to 1,500 cancer patients and survivors reveal that:

  •  the proportion of patients who can recall discussions with their doctors is pretty consistent under a 3 year time frame; failure to recall is more likely when date of last treatment is over 3 years
  •  recall varied by topic - most patients remembered discussions about the need for regular follow-up care, but recollections of discussions about long-term side effects or lifestyle/health recommendations was less consistent over time
  •  effects of time lapse on discussion recall remained significant after controlling for socio-economic attributes
  •  nonetheless, main effects demonstrating positive benefits from doctor discussions stayed strong and were not moderated by time lapse since last cancer treatment 

In short, even though time lapse since last treatment does lower chances of patient recall, substantive multivariate model results were not compromised or altered even when the survey sample includes cancer survivors who had received their last treatment over 3 years ago.


Background

This cancer patient experiences survey was designed by a consortium led by the National Cancer Institute, in collaboration with the American Cancer Society (ACS), the Centers for Disease Control and Prevention (CDC), the NIH Office of Behavioral and Social Sciences Research (OBSSR), LIVESTRONG, and the Agency for Healthcare Research and Quality (AHRQ). The survey was a self-administered paper-and-pencil questionnaire fielded to cancer patients and survivors during Panel 15 Round 5 and Panel 16 Round 3 of the 2011 Medical Expenditure Panel Survey (MEPS*). 

Because the survey was administered at only one point in time, it captured perceptions of cancer patients and survivors in vastly different stages of cancer treatment and recovery. Some respondents were currently receiving treatment for cancer, while others had their last cancer treatment over 10 or 20 years ago. This wide range in time frames provided an opportunity to compare recall of doctor-patient communications by  time lapse since last treatment.

pct of respondents by date of last cancer treatment.jpg

Patient Recall is a Subjective Measure

No one could dispute the fact that patient recall may not reflect what was truly discussed by the doctor. The value of patient research is to capture patient perceptions and experiences, from the perspective of the patients. In essence, the focus of this analysis is on the differences between patients who recalled vs. did not recall discussion of specific topics, not what their doctors truly did discuss. 


Patient Failure to Recall by Time Lapse since Last Treatment

All respondents were asked if their doctor had ever discussed:

  •  need for regular follow-up care and monitoring after completing cancer treatment
  •  long-term side effects of cancer treatment
  •  emotional or social needs related to cancer
  •  lifestyle or health recommendations during and after completing cancer treatment

On each of these questions, respondents could select the option “Did Not Remember” if they could not recall whether their doctor has discussed that particular topic with them. Hence, the dependent variable was coded 1 for respondents who explicitly indicated that they did not remember whether their doctor discussed that topic; and 0 for all other respondents. Results on these binary recall variables are shown below. 

The red line represents proportion of respondents who could NOT remember whether their doctor had ever discussed with them the need for regular follow-up care and monitoring after completing cancer treatment - this proportion stayed very low at under 5% for all current patients and survivors who received their last treatment for cancer less than 5 years ago. Similarly, the proportion of respondents who could NOT remember whether their doctor had ever discussed long-term side effects (green line) or made lifestyle or health recommendations (purple line) stayed around or under 5% for all current patients and survivors who received their last treatment for cancer less than 3 years ago. The only exception appears to be discussion of emotional or social needs related to cancer - failure to recall jumped to 10% or more when last treatment date exceeded one year. Taken together, these effects suggest that the proportion of patients who can recall discussions with their doctors is pretty consistent under a 3 year time frame.

patient failure to recall by date of last cancer treatment.jpg

Binary logistic models confirmed statistically significant linear effects of date of last treatment on likelihood to remember, such that survivors who were treated > 20 years ago were:

  •  6.3 times less likely to remember whether doctor ever discussed need for regular follow-up care and monitoring 
  •  5.1 times less likely to remember whether doctor ever discussed long-term side effects of cancer treatment 
  •  3.7 times less likely to remember whether doctor ever discussed emotional or social needs related to cancer 
  •  5.9 times less likely to remember whether doctor ever discussed lifestyle or health recommendations 

compared to cancer patients who are currently receiving treatment.

These odds ratios stayed consistently high even after controlling for socio-economic attributes such as household income, education, age, gender, race, and whether patient had insurance coverage for cancer treatment.

Patient Recall of Discussions by Time Lapse since Last Treatment

Among the cancer patients and survivors who could remember whether or not their doctor had ever discussed these topics, the proportion who recalled a discussion took place varied widely by topic. As shown below, the red line indicates that for all survivors who received treatment less than 20 years ago, over 90% affirmed that their doctor discussed the need for regular follow-up care and monitoring after completing cancer treatment. In contrast, the proportion of respondents who recalled their doctor discussing long-term side effects (green line) or lifestyle or health recommendations (purple line) was significantly lower among survivors who were no longer receiving treatment for cancer. Among the 4 topics, all patients were least likely to recall their doctors ever discussing emotional or social needs related to cancer; but relatively, it was clear that proportion of respondents who recalled having such a discussion was significantly lower among survivors who were no longer receiving treatment for cancer compared to current cancer patients.

patient recall of discussion by date of last cancer treatment.jpg

Compared to the odds ratios above on likelihood to recall, the effect of date of last treatment on whether respondents said a discussion ever occurred was relatively weak - ranging between 1.3 times to 1.7 times more likely to say discussion took place. Again, these odds ratios remained the same after controlling for socio-economic attributes such as household income, education, age, gender, race, and whether patient had insurance coverage for cancer treatment.

Moderating Impact of Doctor Communications on Long Term Well-being

One long term objective of the doctor communications is to minimize stress and anxiety, and support patients’ subjective well-being post-treatment. All respondents were asked to how often they worried that their cancer may come back or get worse on a 5-point scale ranging from “never”, “rarely”, “sometimes”, “often”, and “all the time”. This 5-level variable served as the dependent variable in an ANOVA assessing the main effect of each discussion topic on patients' subjective well-being. Results confirmed that patients and survivors worried less if their doctor had ever discussed:

  •  need for regular follow-up care and monitoring after completing cancer treatment (F=12.07, p<0.001)
  •  long-term side effects of cancer treatment (F=24.90, p<0.001)
  •  emotional or social needs related to cancer (F=4.18, p<0.05)
  •  lifestyle or health recommendations during and after completing cancer treatment (F=20.86, p<0.001)
Positive Impact of Doctor Discussion on Patient Well being.jpg

Clearly, such discussions benefit patients’ subjective well-being. The question is whether the evident benefits from such discussions are reduced over time? Thus, the ANOVA model was expanded to include interactions between date of last cancer treatment and each discussion topic to examine whether the length of time that had passed since the discussion might moderate the benefits of each discussion topic. Most interaction terms emerged as null (p>0.50) except for interaction between time lapse and discussion on emotional or social needs related to cancer (F=2.28, p<0.05). Given that only 1 out of 4 interaction terms emerged as significant, it is possible that the moderator effect of time was not due to impaired recall, but due to an actual trend that doctors were less likely to discuss emotional or social needs related to cancer decades ago compared to recent years. Hence, for the most part, the substantive model results are not compromised if the survey sample includes cancer survivors who had received their last cancer treatment many years ago.


Methodology Notes: Instead of the MEPS 2011 person weight, a sub-sample weight (CSAQW11F) was created specifically to be used with the Cancer SAQ for persons who were age 18 and older at the interview date. This weight adjusts for the cancer SAQ nonresponse and reflects the adult population self-reporting as having been diagnosed with or treated for cancer as an adult. 

* The Medical Expenditure Panel Survey (MEPS) is a nationwide survey conducted by the U.S. Agency for Healthcare Research and Quality (AHRQ). The sample is designed to represent the general U.S. population of all ages, and is arguably the most reliable source of data on Americans’ health care utilization and expenditure.