Noncompliance to guidelines in head and neck cancer treatment; associated factors for both patient and physician

Subjects

This retrospective study, based on medical records, included patients with newly diagnosed
HNSCC without distant metastasis. Patients with cancer of the lip, oral cavity, nasopharynx,
oropharynx, hypopharynx, and larynx which could be treated with curative intent qualified
for this study. Recurrent or residual cancer was excluded but patients with second
primary HNSCC were deemed eligible. Patients who were enrolled in any clinical trial
in this period were also excluded. In the period from January 2010 to December 2012,
829 patients were included. The study was carried out in compliance with the Helsinki
declaration and was approved by the ethics committee of the Erasmus Medical Center,
including a waiver for informed consent.

All patients were initially set for curative treatment at the Erasmus Medical Center
Rotterdam, the Netherlands. The tumor stage at the time of first diagnosis was classified
according to the clinical staging system described by the Union for International
Cancer Control (UICC). A first treatment proposal was presented at the regional multidisciplinary
head and neck tumor conference, where all new patients were discussed. The multidisciplinary
tumor board (MDT) consisted of oncologists, head and neck surgeons, and radiotherapists.
The treatment proposal was weighed up against the standard treatment protocol, which
is based on national guidelines published by the Comprehensive Cancer Centre the Netherlands
(IKNL) and regional additions. The final proposal may be according to the guidelines
(standard treatment) or deviant (nonstandard treatment). Reasons for nonstandard treatment,
either as a result of MDT or patient decision, were collected retrospectively. Solely
major deviations of standard guidelines were marked as ‘nonstandard’ treatment. A
change in dose of radiotherapy or chemotherapy was not accepted as a deviation of
standard guidelines, but refusing total laryngectomy indeed was.

Outcomes

Following the discussion in the MDT, the treatment proposal was discussed with the
patient. In the decision making process, patients may have either accepted or declined
the proposal. In this study, we considered the following groups.

1. Standard treatment according to guidelines (reference group)

2. Nonstandard treatment as proposed by the multidisciplinary tumor board

3. Nonstandard treatment as desired by the patient:

a. Alternative (less extensive)

b. No treatment at all

Different parameters present at the time of diagnosis, were retrospectively collected
for every patient. These included age at diagnosis, year of diagnosis, tumor site,
tumor stage, gender, marital status, having children, comorbidity conditions, prior malignancy
(head and neck or other), treating physician (head and neck oncologist, radiotherapist
or general oncologist) and survival. The presence of one or more different comorbid
ailments was coded for all patients using Adult Comorbidity Evaluation-27 (ACE-27)
7]. The ACE-27 grades specific comorbid conditions in different organ systems into one
of three levels of comorbidity. The overall comorbid score is graded in four levels,
none, mild, moderate or severe and is based on the highest ranked single ailment.
Patients with two or more moderate ailments in different organ systems or disease
groupings are graded as severe. The ACE-27 is a comprehensive tool, commonly used
in head and neck cancer literature, and accurate as a retrospective measuring instrument
of comorbidity.

The retrospective analysis of the specified characteristics was performed by the first
two authors (EACD an SWM) who were not involved in decision making by the multidisciplinary
tumor team.

Statistical analysis

The data was analyzed with IBM SPSS Statistics version 21.0 for Windows. For statistical
processing, several variables were converted to dichotomous values, based on experience,
evidence from literature, or distribution of data following a normal Gaussian curve
with a cutoff point at the mean. This was the case for age (65 or ?65 years), marital
status (partner or single), comorbidity (ACE-score 0–1 or ACE-score 2–3), tumor site
(pharynx, larynx and oral cavity) and tumor stage (stage I-II or stage III-IV). Descriptive
statistics,?
²
tests and simple logistic regression methods were used to compare three groups (reference
group, nonstandard treatment by MDT decision and nonstandard treatment by patient’s
decision) P-values 0.05 were considered to be statistically significant.

Multivariate analysis using logistic regression methods and taking into account interaction
terms was performed to determine predictive factors associated with non-standard treatment
following MDT or patient decision 8]. A predictor was defined as a predictive factor that contributes independently and
significantly (p-value of??0.05) to the choice of non-standard treatment, done either
by the MDT or by patient decision. In general, the limiting sample size in logistic
regression analysis is the number of events of interest. The assumption is made that
this analysis will produce reasonably stable estimates of the effect of each variable
on the outcome if the limiting sample size allows a ratio of approximately 10 to 15
observations per possible predictive factor 8]. To gain insight into the impact of each possible predictor in the model, all variables
were entered in the logistic regression analysis at the same time. The following factors
were included: age at diagnosis, year of diagnosis, gender, marital status, having
children, tumor stage, tumor site , comorbidity, prior malignancy and prior head and
neck malignancy, and type of initial treatment following national guidelines. Stratification
by gender was done following the analysis for interaction terms. To design a final
stratified model showing independently and significantly predictive factors associated
with non-standard treatment following MDT or patient decisions, a backward selection
procedure was applied, accepting predictors with a p-value 0.05. Following this,
a forward selection procedure was done to confirm our results.

To gain insight in survival of the different groups of patients, we applied a Cox
regression analysis. After checking the proportional hazards assumption for each variable,
we adjusted the survival analysis for gender, age, tumor site, tumor stage, comorbidity
and a history of having a prior tumor.