Development of Prognosis in Palliative care Study (PiPS) predictor models to improve prognostication in advanced cancer: prospective cohort study

Both Prof Barrett-Lee and Dr Sundar have described the predictive
power of the PiPS prognostic tool as little better than "tossing a coin".
However, it should be remembered that the instrument is trying to
determine whether patients are expected to live for "days" (less than 14
days), "weeks" (2 - 7 weeks) or "months+" (8 weeks or greater). Since this
represents three distinct prognostic categories one would expect a random
prediction to be correct approximately 33% of the time (not 50%). In our
study we found that a multi-professional prediction of survival was
significantly better than this (being correct on 53.7% of occasions -
Table 6), and that the PiPS-B was correct on 61.5% of occasions Although,
after adjusting for "over-optimism", the PiPS-B score was not
statistically significantly better than a multi-professional estimate of
survival, it was better than either a doctor or a nurse alone and was
certainly better than a random "guess".

As Dr Sundar has noted there were a number of limitations to our
study. The refusal rate of 57% of eligible patients was disappointing, but
not unexpected given the nature of the study population. Unfortunately we
do not know whether the survival of non-participants was different to that
of the participants (it was not deemed to be ethical to "flag" the health
records of the non-participants). However, we found no differences in the
age or site of primary cancer between the participants (n = 1018) and the
eligible patients who were not recruited (n = 5994). There were slightly
more men among the recruited patients than among the non-recruited (54.7%
vs 51.9%, p = 0.09).

It is true that some of the patients in the study only had locally
advanced disease, however we did not collect data regarding the reasons
why these patients were no longer receiving active treatment. The decision
to stop disease-modifying treatment was taken by individual consultant
oncologists and would have been influenced by factors such as poor
performance status, co-morbidities, general frailty, site of primary
disease and previous treatment. The presence of metastatic disease at any
site (and particularly liver metastases) was associated with a worse
prognosis (Table 3 and 4).

The majority of patients included in the PiPS study were recruited
from independent hospices (67.7%), with 17.9% coming from hospital
palliative care teams and 11.9% recruited from community services.
Details regarding the number of previous cancer treatments,the prior use
of trial medications, or disease-free interval prior to development of
metastases were not collected as part of this study. Although this
information is frequently easily available in a hospital setting, it is
sometimes difficult to glean such detailed clinical information from
hospice or community notes.It was beyond the resources of the current
study to track down this type of "missing" information on all
participants. An attempt was made to document the time between diagnosis
and referral to palliative care and the stage at diagnosis, however for
the reasons previously explained this information was poorly recorded and
could not be included in the analysis. Age, gender and co-morbidities were
collected but did not feature as significant variables in the prognostic
models.

Although, as other correspondence on this issue has demonstrated,
there are some patients who would rather not know their prognosis, there
is also evidence that many patients do value such information (Degner,
Kristjanson et al. 1997; Steinhauser, Christakis et al. 2001; Adams,
Boulton et al. 2009). Interestingly, only 6% of the patients approached
about the study refused to consent because they were concerned about
discussing their prognosis. In answer to Dr Sundar's query, we did ask
study participants for their views about receiving prognostic information
and this data will be presented in a subsequent paper. We agree that it is
important to investigate the best way for prognostic information to be
communicated. In our next study we are proposing to undertake in-depth
qualitative work with patients, relatives and healthcare professionals to
determine the clinical utility of the PiPS instruments.

References

Adams, E., M. Boulton, et al. (2009). "The information needs of
partners and family members of cancer patients: a systematic literature
review." Patient Education & Counseling77(2): 179-186.

Degner, L. F., L. J. Kristjanson, et al. (1997). "Information needs
and decisional preferences in women with breast cancer." JAMA277(18): 1485
-1492.

Steinhauser, K. E., N. A. Christakis, et al. (2001). "Preparing for
the end of life: preferences of patients, families, physicians, and other
care providers." Journal of Pain & Symptom Management22(3): 727-737.

The "simple" answer is that our patients often do in fact want an
idea of their life expectancy, so that they can plan ahead for themselves
and their families. Therefore this is an important research question that
is patient centred, and the authors should be commended for attempting to
move knowledge in this difficult area forward. Unfortunately the current
tool is little better than the 50% predictive power of tossing a coin in
statistical terms.

Perhaps a combined model using clinicians' estimates and PiPS could
have been considered.

Whilst both this paper and "Predicting and communicating prognosis in
palliative care" are well-meaning, they exhibit a major failing of the
medical profession on the core issue: enhancing cancer patients' survival
time, rather than the somewhat academic exercise of prediction based on
averages - a reverse focus.

You will have noted that both papers focus on clinical aspects of
prognosis. Neither address the issue of how patients can increase their
survival by attending to various aspects of their lifestyle. This latter
is now known to precisely have this effect. It is way past the time that
this was brought to the fore.

It is not enough to say to a patient "given your age, state of
health, cancer type & stage, etc, then we expect you to live x
weeks/months/years" without adding "however, if you increase your
exercise, reduce your overweight, optimise your Vit.D3 and EFA levels, eat
lots of fresh vegetables and fruit, cut out sugar, reduce alcohol, then
you can signficantly increase the odds of surviving much longer. Not only
this, if you check how you're doing every so often, you can tell if you
are on track."

If any of these 'howevers' were a new drug, their worth would be far
greater than many of those presently used. There is also the oddity of why
huge amounts of money is poured into clinical-based treatments and
virtually nothing into patient controlled lifestyle factors.

Clinicians may feel this is up to the patient or their GP. What is
certain is that most cancer patients are not aware of many of these
'howevers', and/or need motivating (by regular monitoring, at less cost
than most clinical treatments) to do this.

Attending to this would enable the UK to meet its target of saving
5,000 extra cancer patient lives in the next couple of years.

My own experience underlines my recommendations. The first several (3
face-face with me, the others via their case conferences) consultant
experts all said I only had weeks to months to live - 4 years ago. It has
only been my dogged search and immplementation of what I found out that I
have survived so far. No thanks to any of those earlier doctors, none of
whom told me of this research that enabled me to so extend my life.

Yours sincerely, Ian