Keywords:
Advance directives - Medical oncology - Palliative care - Living wills
Descritores:
Diretivas antecipadas - Oncologia médica - Cuidado paliativo - Declaração de vontade
INTRODUCTION
Advance directives (AD) are an essential part of the Advance Care Plan (ACP), which
aims to improve patient care, quality of life, and reduce health costs.[
[1]
] The formulation of an AD involves a behavioral change and a better understanding
of the context of a terminal disease, taking into account the values, beliefs and
objectives of the patient.[
[2]
] The prevalence of advance directives in the USA ranged from 5-37% but this number
can reach 70% in samples affected by terminal conditions.[
[3]
[4]
[5]
[6]
]
The benefits of advance directives include reduction of in-hospital mortality rates,
higher levels of patient satisfaction, the accomplishment of patient's wills, and
reduction of depression and anxiety in end-oflife, for patients and their relatives.[
[7]
[8]
[9]
[10]
[11]
[12]
] The relevance of AD grows exponentially as we look at the current reality of Oncology
and the “chronification” of disease. The process of “chronification” of disease put
the ADs into a risk of being undervalued or postponed. That leads us to a potentially
negative impact with unwanted costs in a collapsing global health system.[
[13]
[14]
[15]
[16]
]
Every oncologist treating incurable cancer patients in daily practice recognizes the
impact of the discontinuation of anticancer treatment and the beginning of the called
Best Supportive Care (BSC) modality of treatment. Despite this, there is a lack of
data regarding the association of discontinuation of antitumor treatment with advance
directives elaboration and implementation. The primary objective of this study is
to evaluate if the moment of discontinuation of antitumor treatment could influence
the rates of advance directives.
METHODS
Study design and patients
We perform a retrospective study involving cancer patients in palliative treatment
at a university hospital in Brazil. Convenience sampling was made by the registry
number of their medical records generated by the hospital's electronic system. Eligibility
criteria included patients over 18-years-old at the time of disease, an incurable
disease (at diagnosis or at progression) and ongoing palliative treatment. Patients
with cognitive impairment or documented inability to make decisions were excluded.
For data analysis, the patients were divided into two groups, one with and another
without advance directives. The cases were followed from the time of diagnosis of
the incurable condition until death.
We select the first living will declaration directly manifested by the patients and
documented in their medical record as the advance directive to consider in the analysis.
This could refer to any will regarding the place of permanence, place of death, and
treatment measures as cardiopulmonary resuscitation, mechanical ventilation, chemotherapy,
enteral tubes, dialysis, broad-spectrum antibiotics and blood transfusions.[
[17]
[18]
[19]
] Demographic characteristics were compared between the two groups. Clinical variables
included tumor primary site, clinical stage at diagnosis by TNM 7th ed.,[
[20]
] performance status by ECOG (Eastern Cooperative Oncology Group) classification at
diagnosis, age, sex, educational level, marital status, palliative antitumor treatment
dispended, intensive care unit admissions, time intervals from the incurable diagnosis
to the discontinuation of the antitumor treatment and beginning of Best Supportive Care. The final outcomes included the medical specialty in the last medical contact, cause
of death, place of death, life- sustaining measures received and time intervals to
the limitation of therapeutic resources (also called Do-Not-Resuscitate order ) and death.
As a definition of Best Supportive Care (BSC) we consider any palliative treatment,
excluding antineoplastic treatments.[
[21]
[22]
] As a definition for the limitation of therapeutic resources, we used the documentation
in medical records of the contraindication to aggressive measures, something equivalent
to do-not-resuscitate orders.
The study was analyzed and approved by the Ethics and Research Committee of our Institution
and is registered in the Brazilian National Database for Scientific Research (Plataforma
Brazil) with the number CAAE 71559817.6.0000.5327.
Data collect
Data were collected retrospectively from electronic medical records by direct analysis
of documented information. For patients dying outside the hospital and no death registry
in the electronic system, an active search was conducted by telephone contact to assess
the outcome details. The patients included in this historical cohort were selected
from the attendances realized between 2013-2017 and the data were analyzed between
October 2017 and June 2018.
Statistical Analysis
Quantitative variables were described by mean and median. The distribution of the
qualitative variables was analyzed using percentages. The statistical analysis was
performed with SPSS v.18 using the t -test for continuous variables and the chi-square test for categorical variables.
The two groups were compared using uni and multivariable logistic regression. The
performance status was dichotomized in ECOG 0-3 vs ECOG 4. Places, where the active treatment was discontinued, were also dichotomized
in outpatient scenario (mainly doctor's office) and inpatient setting (including the
emergency department, wards and ICU). An adjusted Poisson's regression analysis was
done for place and performance status at the moment of the discontinuation of anticancer
treatment. To determine the sample size we calculated at least 300 patients estimating
approximately 25% prevalence of advance directives, providing 80% power to detect
a 2.0 hazard ratio for the association between AD and other variables, with a significance
level of 5%. We added 10% of possible losses to reach the final sample size. The missing
data were managed with pairwise deletion. A p -value below 0.05 was considered statistically significant.
RESULTS
A total of 390 electronic medical records were accessed and 321 subjects confirmed
all eligibility criteria and were included in the analysis. The demographic characteristics
of the groups with and without advance directives were balanced and summarized in[Table 1]
. The prevalence of advance directives was 22.7% and 74% of subjects received at least
one cycle of antitumor treatment, a high rate based on literature evidence.[
[23]
]
Table 1
Characteristics of Patients with and without advance directives.
|
Total 321 pts[*]
|
Without Advance Directives
|
With Advance Directives
|
P-value
|
|
248(77,2%)
|
73(22,7%)
|
|
Age (median)
Gender
|
68
|
69
|
0,946[a]
|
|
Female
|
138(55,6%)
|
35(47,9%)
|
0,972[a]
|
|
Male
|
110(44,4%)
|
38(52,1%)
|
|
Years in school
|
|
|
|
|
< 8 years
|
134(54%)
|
39(53,4%)
|
|
|
8 - 11 years
|
63(25,4%)
|
19(26,0%)
|
0,977[a]
|
|
> 11 years
Marital Status
|
51(20,6%)
|
15(20,5%)
|
|
|
Married
|
152(61,3%)
|
38(52,1)%
|
|
|
Single
|
76(30,6%)
|
28(38,4%)
|
0,366[a]
|
|
Divorced
|
20(8,1%)
|
7(9,6%)
|
|
|
Primary Site of the Tumor Upper Gastrointestinal Tract
|
55(22,2%)
|
15(20,5%)
|
|
|
Lung
|
48(19,4%)
|
15(20,5%)
|
|
|
Breast
|
23(9,3%)
|
6(8,2%)
|
0,991[a]
|
|
Lower Gastrointestinal Tract
|
46(18,5%)
|
13(17,8%)
|
|
|
Others
|
76(30,6%)
|
24(32,9%)
|
|
|
Clinical Stage at Diagnosis
1
|
7(2,8%)
|
2(2,7%)
|
|
|
2
|
34(13,7%)
|
8(11%)
|
|
|
3
|
67(27,0%)
|
18(24,7%)
|
0,993[a]
|
|
4
|
140(56,5%)
|
45(61,6%)
|
|
|
PS ECOG at Diagnosis[*]
0
|
38(15,7%)
|
12(16,7%)
|
|
|
1
|
113(46,7%)
|
38(52,8%)
|
|
|
2
|
51(21,1%)
|
10(13,9%)
|
0,593[a]
|
|
3
|
35(14,5%)
|
12(16,7%)
|
|
4
|
5(2,1%)
|
0
|
|
|
Palliative Chemotherapy
Yes
|
184(74,2%)
|
54(74%)
|
0,970[a]
|
|
No
|
64(25,8%)
|
19(26,0%)
|
|
BSC registry[**]
Yes
|
191(77,6%)
|
64(87,7%)
|
0,077[a]
|
|
No
|
55(22,4%)
|
9(12,3%)
|
|
PS ECOG at BSC#
|
|
|
|
|
0 - 3
|
112(58,6%)
|
47(74,6%)
|
0,137[a]
|
|
4
|
79(41,4%)
|
16(25,4%)
|
|
Place at BSC#
|
|
|
|
|
Outpatient
|
123(64,4%)
|
52(82,5%)
|
0,043[a]
|
|
Inpatient+
|
68(35,6%)
|
11(17,5%)
|
a Wald Chi-square test.
* Total n=314 pts, due to 7 missing data.
** Total n=319 pts, due to 2 missing data (Total patients with BSC registry: 255 due
to 11 missing values of AD). # 254 patients due to 1 missing information of PS ECOG
and Place at BSC. + Inpatient includes Emergency department, clinical wards, and Intensive
Care Unit (ICU). PS ECOG, Eastern Cooperative Oncology Group Scale of Performance
Status. BSC, Best Supportive Care.
The group with AD correlated more with the registry of treatment discontinuation compared
versus the no AD group (87.7% vs 77.6%) and temporal correlation analysis found that the treatment discontinuation
decision occurs on the same day or before the first AD manifestation in 82.3% of cases.
Looking at the moment of treatment discontinuation, the place at this moment seems
to correlate in a different way between the groups with and without ADs, as demonstrated
in[Figure 1]
. The group with advance directives was more correlated with the outpatient scenario
than the group without advance directives (82.5% vs 64.4%, respectively; RR 1.88; 95%CI 1.019-3.496; p =0.043) by multivariate regression. The same analysis did not show a significant
association between ECOG performance status (0-3 vs 4) and the two groups in the moment of BSC beginning ( p =0.137).
Figure 1 Places at discontinuation of active treatment and initiation of Best Supportive Care. This graphic illustrates the differences of scenarios between the groups with and
without advance directives when Best Supportive Care was initiated.
An analysis, independent of the advance directive status, was performed to access
the correlation between performance status and place where BSC was implemented. A
statistically significant correlation was found between better performance status
(PS ECOG 0-3) and the outpatient scenario in comparison with hospitalized patients
(RR 1.835; 95% CI 1.387-2.429; p <0.0001), as demonstrated in[Figure 2].
Figure 2 Distribution of patients by place and performance status at the moment of BSC decision
(n=266). This graphic illustrates the correlation of outpatient scenario with better
performance status scores. ECOG: Eastern Cooperative Oncology Group. Inpatient: Emergency,
intensive care unit and wards included.
There was no statistical difference in outcomes between the groups with and without
an AD. The outcomes included the limitation of therapeutic resources, place of death,
receipt of blood components, ICU admissions, invasive palliative procedures (drainage
of cavities, biliary tract and derivation of intestinal transit in the majority),
the specialty at last assistance, cause of death, orotracheal intubation, mechanical
ventilation, artificial parenteral hydration and enteral nutrition, and use of broad-spectrum
antibiotics did not differ between groups with and without AD ([Table 2]). The time interval between the palliative diagnosis and the start of BSC, limitation
of therapeutic resources, and death also did not differ between the two groups. The
median interval from palliative diagnosis to death differed by 3 days between the
groups with and without advance directives. The most cited first living will was the
“desire to die at home”, representing almost 30% of the registries. Among those patients
that reported this desire, 13.6% died at home, compared with 64% of inhospital deaths.
Table 2
Outcomes of patients with and without advance directives
|
Outcomes
|
Wthout advance directives
|
With advance directives
|
P-value
|
|
Place at Limitation of Therapeutic resources (n=267[*] )
|
201(100%)
|
66(100%)
|
|
|
Clinical Wards
|
96(47,8%)
|
23(34,8%)
|
|
|
Emergency
|
53(26,4%)
|
21(31,8%)
|
0,230a
|
|
Doctor's office
|
44(21,9%)
|
19(28,8%)
|
|
ICU
|
7(3,5%)
|
3(4,5%)
|
|
|
Place of death(n=293[*] )
|
232(100%)
|
61(100%)
|
|
|
Palliative Unit in the Hospital
|
109(47%)
|
31(50,8%)
|
|
|
Emergency
|
43(18,5%)
|
9(14,8%)
|
|
|
Clinical Wards
|
38(16,4%)
|
11(18%)
|
0,905a
|
|
Home
|
25(10,8%)
|
5(8,2%)
|
|
Other
|
10(4,3%)
|
2(3,3%)
|
|
|
ICU
|
7(3%)
|
3(4,9%)
|
|
|
Patients that received Blood transfusions
|
136(54,8%)
|
41(56,2%)
|
0,841a
|
|
Patients with ICU admission
|
14(5,6%)
|
5(8,3%)
|
0,509a
|
|
Patients that received Palliative
Procedures
|
48(20,3%)
|
17(25%)
|
0,409a
|
|
Last assistance specialty (n=260[*] )
|
204(100%)
|
56(100%)
|
|
|
Medical Oncology
|
101(48,7%)
|
26(46,4%)
|
|
|
Emergency
|
42(20,2%)
|
9(16,1%)
|
|
|
Palliative Care
|
30(14,7%)
|
13(23,2%)
|
0,738a
|
|
Internal medicine
|
23(11,3%)
|
6(10,7%)
|
|
Intensive Care
|
7(3,4%)
|
2(3,6%)
|
|
|
Other
|
1(0,5%)
|
0
|
|
|
Cause of Death (n=255[*] )
|
200(100%)
|
55(100%)
|
|
|
Progression of Disease
|
163(81,5%)
|
47(85,5%)
|
|
|
Infectious
|
28(14%)
|
7(12,7%)
|
0,396a
|
|
Vascular
|
6(3%)
|
1(1,8%)
|
|
Other
|
3(1,5%)
|
0
|
|
|
Limitation of therapeutic resources
Yes
|
195(79,6%)
|
65(89%)
|
0,085a
|
|
No
|
50(20,4%)
|
8(11%)
|
|
Life-sustaining measures
Cardiopulmonary resuscitation
|
4/231(1,7%)
|
1/61(1,6%)
|
0,961a
|
|
Invasive Airway
|
11/231(4,8%)
|
3/61(4,9%)
|
0,960a
|
|
Parenteral Hydration
|
144/221(65,2%)
|
36/59(61%)
|
0,555a
|
|
Artificial Enteral Nutrition
|
55/220(25%)
|
17/58(29,3%)
|
0,505a
|
|
Broad spectrum antibiotics
|
103/222(46,3%)
|
30/59(50,8%)
|
0,543a
|
|
Interval between Palliative Diagnosis and Death (median in days)
|
464
|
461
|
0,296b
|
(A) Pearson chi-square test; (B) Log Rank (Mantel-Cox) Test
* Varied values due to missing data in medical records for each outcome ICU, Intensive
Care Unit.
DISCUSSION
The positive correlation between the group with advance directives and the outpatient
scenario in the moment of discontinuation of anticancer treatment could be explained
by a greater sense of security, confidence and stronger relation with the assistant
professional.[
[24]
[25]
] These factors facilitate the transition to a BSC approach and also improve the chance
of discussions about the advance care plan and advance directives providing patients
with more autonomy. The emergency, wards and ICUs environment interfere negatively
in the compliance and elaboration of an advance care plan and advance directives.
A good and early advance directive can reduce visits to the emergency department (as
illustrated by the 14.8% vs 18.5% deaths in the emergency), and could also reduce stress and possible iatrogenic
measures as stated by the literature.[
[26]
[27]
[28]
]
Despite no correlation founded between the performance status and ADs documentation,
in daily practice, we observe that the more debilitated patients are less capable
to discuss advance care plan and manifest their advance directives. The prognostic
impact that poor performance status carries may reinforce mercy feelings by oncologists
and/or family members.[
[29]
] A deleterious practice called collusion may occur in this scenario and it refers
to the attitude of put aside the patient's autonomy and restricts the decision process
to relatives and medical team, hiding the truth and facts from the major interested
person.[
[30]
]
A fact that suggests the presence of collusion in this study population is that despite
low rates of ADs, low rates of life-sustaining measures were verified. We also know
that patients on PS ECOG 4 also have more cognitive dysfunctions and weaknesses that
may compromise their interest and ability to formulate advance directives in time
to honor their living wills.[
[31]
] The independent analysis correlating patients in PS ECOG £ 3 with the outpatient
scenario in comparison with PS ECOG 4 (RR 1.835; p <0.0001) corroborates this argument.
The 22.7% prevalence of advance directives found in our study is below that described
in North America but is consistent with data from other localities.[
[3]
[4]
[5]
[6]
[32]
[33]
[34]
] We highlight some possible explanations as a low educational level in our sample
(more than 50% without the basic educational level completed), the recent history
of use of ADs in Brazil, its low dissemination in the health system, and the information
bias inherent to retrospective studies.
Where the oncologic patients have died is an important topic. A large cross-national
study involving people with cancer found a large variety of in-home deaths ranging
from 12-57%.[
[35]
] In this study, 13% of patients died at home despite being the most cited first directive
recorded. Of the 22 patients who registered a directive expressing the will to die
at home only three (13.6%) actually did, compared to fourteen (63.3%) dying in the
hospital. This data clearly indicates the difficulty in honoring this type of living
will, suggesting the multifactorial influence acting on the final results. One of
the multiple factors for this discrepancy is the low prevalence of outpatient hospice
care the Brazilian public health system, a factor that is correlated with more deaths
in the setting of choice.[
[36]
] Other factors may include losses of ADs in the transition between different health
teams, lack of a standardized AD registry and inefficient ways to keep it easily accessible.[
[37]
[38]
]
Although the literature indicates a correlation of the advance directives with a reduction
in the adoption of aggressive and life-sustaining measures in the terminal setting,
in this study the presence of living wills did not statistically correlate with less
life-sustaining treatments.[
[39]
] This could be due to underestimation of AD caused by discussion without adequate
documentation, difficulty in talking with patients about therapeutic limitations leading
to relative-guided decisions, and simply beliefs by family members and physicians
that some life-sustaining measures could have a positive impact on the lifetime of
a terminally ill patient, generating the high rates of parenteral hydration and broad-spectrum
antibiotics. Finally, in accordance with literature evidence, our data show no difference
in the survival curves between the groups with and without advance directives.[
[40]
[41]
]
CONCLUSION
We must consider that, like any observational study, this research is subject to observational
study biases and the results serve as a hypothesis generator. Although it is not able
to directly establish a causal relationship, it provides a statistically significant
correlation between advance directives registration and the outpatient setting when
the anticancer treatment was discontinued, something based on the theoretical rationale.
This type of observational data brings the possibility to change practice improving
care of palliative cancer patients by discussions and elaboration of advance directives,
patients the opportunity to have their wills honored. Finally, we should always keep
in mind the multifactorial pattern that influences the composition of advance directives
and impaired solid base evidence in this field of research. ([Figure 1]).
Advanced directives in Brazil, Latin America and all around the world need to be further
studied from a clinical perspective, in addition to the legal and psychological view.
More prospective large-scale studies are needed to add data to the literature, including
intervention studies. The present study is the first to demonstrate that the place
of discontinuation of antitumor treatment affects the advance directives rates, and
is the first study assessing the reality of advance directives amongst oncologic patients
in Brazil.
This conclusion is important if we consider the current development of the therapeutic
arsenal and the prolongation of patient survival, with a concomitant gain in quality
of life and “chronification” of disease. It brings a natural tendency to postpone
the discussion of the advanced care plan to a later moment after all therapeutic lines
had failed. The transition to best supportive care could represent a window for discussions
about end of life, for the valorization of patient's autonomy, selfdetermination,
and to preserve dignity in the death and dying process. In addition, this study suggests
that the outpatient scenario is the best for the elaboration of advance directives.
Medical assistants should always discuss the aspects of the end of life care with
their patient, preferably in the office and on the most appropriate occasions. The
focus should always be on dignity and respect for the desires of the patient, from
the beginning to the end.
Bibliographical Record
Pedro Grachinski Buiar, Jose Roberto Goldim, Vania Naomi Hirakata. The discontinuation
of antitumor treatment and the advance directives in cancer patients. Brazilian Journal
of Oncology 2020; 16: e-20200006.
DOI: 10.5935/2526-8732.20200006
