Prereferral Head and Neck Cancer Treatment

Abstract

Objective

To evaluate the prereferral treatment of patients referred to our tertiary care center with recurrent or persistent head and neck cancer for compliance with National Comprehensive Cancer Network (NCCN) guidelines.

Design

A prospective recruitment and retrospective chart review.

Patients

The study included new patients identified at multidisciplinary treatment planning conference from October 1, 2008, to February 1, 2009, who had received prior treatment at an outside institution and presented to our department with recurrent or persistent disease.

Main Outcome Measures

All facets of prior care were examined, including the time from initial symptoms to diagnosis and whether their prereferral treatment was compliant with or deviated from NCCN guidelines for head and neck cancer.

Results

A total of 566 consecutive new patients were identified, of whom 107 (18.9%) had persistent or recurrent disease. The average time from first presentation with initial symptoms to diagnosis among patients who presented with persistent disease was 23.8 weeks. Nearly half of the patients who presented with persistent or recurrent disease had either endocrine (21.5%) or cutaneous (24.2%) primary cancers, with the rest of the cases being distributed among 10 other sites. Of the patients who presented with recurrent or persistent disease, 43.0% had prereferral care that was noncompliant with NCCN guidelines. Of these patients, 58.7% had inadequate surgical management, 15.2% were treated for the wrong diagnosis, 10.9% received inadequate adjuvant therapy, 4.4% received inadequate radiotherapy, and 10.9% refused indicated recommended treatment.

Conclusions

Significant deviation from NCCN guidelines for head and neck cancer treatment was observed in the cohort of study patients. The failure to administer adjuvant therapy when indicated by NCCN guidelines is particularly concerning. Economic and noneconomic costs, including lost wages, cost of “do-over” therapy, and potentially diminished survival, are substantial. Measures to ensure that patients receive therapy according to guidelines should be a national priority.

Evaluating the quality of medical care has become increasingly important, as highlighted by the Institute of Medicine’s (IOM) report, “Crossing the Quality Chasm: A New Health System for the 21st Century.”¹ This publication articulated the need to avoid both overuse of ineffective care and underuse of health care practices with proven benefit.² Significantly, the IOM’s report recognized the importance of optimizing the quality of cancer care.³

Most commonly, the outcomes of mortality, morbidity, and length of hospital stay are used as indirect indicators of quality of care. Another approach is to evaluate adherence to clinical practice guidelines as an indicator of the quality of care. Clinical practice guidelines are consensus- and evidence-based recommendations that establish the standard of care that is to be achieved. Also, these guidelines create priorities for quality improvement and validate the indicators by which quality is measured.⁴ Compliance with established clinical practice guidelines has been reported in both academic and community health care settings with regard to carotid endarterectomy,⁵ colonoscopy,^6,7 axillary lymph node dissection,⁸ colorectal cancer care⁹ and surveillance,¹⁰ and management of papillary thyroid cancer¹¹ and acute coronary syndrome,¹² among others.

The National Cancer Comprehensive Network (NCCN) has invested extensive resources to improve the care of patients with head and neck cancer.¹³ To our knowledge, compliance with NCCN guidelines for head and neck cancer outside of tertiary care centers has not been assessed. To determine the quality of care received by head and neck cancer patients in the community setting, we evaluated the prereferral treatment received by patients referred to our tertiary care center with persistent or recurrent disease for compliance with NCCN guidelines.

METHODS

New patients who are referred to the Department of Head and Neck Surgery at the University of Texas MD Anderson Cancer Center, Houston, are routinely presented at a weekly multidisciplinary treatment planning conference attended by head and neck surgeons, radiation oncologists, medical oncologists, dental oncologists, speech pathologists, radiologists, and pathologists. Treatment recommendations are in compliance with NCCN or institutional guidelines whenever possible. All eligible patients are also offered treatment under clinical protocols approved by our institutional review board. New patients were prospectively recruited as they were presented at the conference between October 1, 2008, and February 1, 2009. Their prereferral medical records were subsequently retrospectively reviewed. For those patients who had previously been treated elsewhere and who were presenting with persistent or recurrent disease, this included a review of available outside records detailing previous imaging and pathology reports, treatment, and office visits. All cases that presented with persistent or recurrent disease were reviewed by the first (C.M.L.) and senior (R.S.W.) authors, and a consensus was reached. For each review, prereferral treatment was correlated with NCCN guidelines to determine compliance or deviation. Outside records are diligently collected by our business center, which contacts both the patient and the referring physician to obtain all existing records regarding prereferral care. Despite our best efforts, obtaining complete outside records, including documentation of all patient visits, was difficult. Among patients who presented with persistent or recurrent disease, complete records were available for 66.7% of those evaluated by medical or radiation oncologists, half of those evaluated by general surgeons, 7.5% of those evaluated by dermatologists, 6.1% of those evaluated by otolaryngologists, and 4.6% of those evaluated by primary care physicians. The rest of the patients’ treatment histories were obtained from patient interview, pathology and imaging reports, and operative records.

In addition to demographic data, collected information included time to diagnosis, outside provider information, and details of prereferral workup and treatment. For patients who presented with persistent or recurrent disease, an assessment of compliance with NCCN guidelines was made based on available prereferral information by the first (C.M.L.) and senior (R.S.W.) authors. This evaluation included a rigorous assessment of available operative notes, summaries of adjuvant therapy, histopathology and imaging reports, and clinic notes for each patient.

Whether surgical treatment was insufficient was determined by a review of all prereferral imaging and pathology reports and operative records. For example, 1 patient underwent an indicated neck dissection, but the corresponding pathology report did not identify any lymph nodes in the surgical specimen. This patient was deemed to have received inadequate surgical treatment. Likewise, patients with lymphadenopathy on preoperative imaging that was still present on postoperative imaging fell into this category. Inadequate primary radiation therapy refers to cases in which patients did not receive adequate dosing, did not complete treatment, or had unexplained or extended treatment breaks. Assessment was based on radiation oncology summaries and clinic notes. The adequacy of adjuvant treatment was also determined by relevant office notes and summaries. Examples of patients who were treated for the wrong diagnosis include those with metastatic neck masses who were treated with antibiotics for more than 6 months before being diagnosed as having malignancy.

Commercial statistical software was used to enumerate descriptive statistics for scaled values and frequencies of study patients within the categories for each of the parameters of interest. Correlations between parameters and end points were assessed by Pearson χ² test or, in instances in which there are fewer than 10 subjects in any cell of a 2×2 grid, by 2-tailed Fisher exact test. Intervals between initial presentation and diagnosis were compared with the Kruskal-Wallis analysis of variance by rank multiple comparisons test. These calculations were performed with a commercially available statistical software package (Statistica; StatSoft Inc, Tulsa, Oklahoma).

RESULTS

A total of 566 consecutive new patients (355 men [62.7%] and 211 women [37.3%]) were identified at multidisciplinary treatment planning conferences between October 1, 2008, and February 1, 2009. Demographic data are summarized in Table 1. Briefly, for the entire cohort, the mean age at presentation was 56.6 years (median age, 57.4 years; age range, 7.5–92.8 years). Four hundred forty-one patients (77.9%) were white, 57 (10.0%) were Hispanic, and 49 (8.6%) were black. Otolaryngologists accounted for 38.7% of referring physicians, and 25.0% of patients were self-referred. The rest of referrals were made by primary care physicians (11.7%), dermatologists (7.2%), endocrinologists (5.2%), dental professionals (27%), oncologists (3.8%), and general surgeons (1.4%). Table 1 also details the subsites of disease on presentation.

Table 1.

Patient Demographics

Variable	No. (%) of Patients
Age at presentation, median (range), y	57.4 (7.5–92.8)
Sex
Male	355 (62.7)
Female	211 (37.3)
Race/Ethnicity
White	441 (77.9)
Hispanic	57 (10.1)
Black	49 (8.6)
Asian	18 (3.2)
Other	1 (0.2)
Referring physician
Otolaryngologist	215 (38.7)
Primary care	65 (11.7)
Dental professional	27 (4.9)
Endocrinologist	29 (5.2)
Oncologist	21 (3.8)
Dermatologist	40 (7.2)
General surgeon	8 (1.4)
Self-referral	139 (25.0)
Other	11 (1.4)
Subsite of disease
Oral cavity	70 (14.2)
Oropharynx	84 (17)
Hypopharynx	10 (2.0)
Larynx	50 (10.1)
Nasopharynx	9 (1.8)
Endocrine	93 (18.9)
Skin	88 (17.8)
Salivary gland	25 (5.1)
Mandible/maxilla	7 (1.4)
Paranasal sinuses	9 (1.8)
Soft tissue	5 (1.0)
Eye	2 (0.4)
Unknown primary	17 (3.4)
Other	13 (2.6)

The majority of patients (65.3%) presented for initial workup and treatment; 13.5% presented with recurrent disease and 5.5% with persistent disease. Patients with no evidence of disease who were transferring their surveillance care accounted for 6.4% of cases, and those who were referred for completion of treatment (generally for postoperative adjuvant therapy) accounted for 6.8% (Table 2). The median time from initial presentation to diagnosis was 3.9 weeks (range, <0.1–170.4 weeks) for patients who were referred for initial treatment, 0.3 weeks (range, <0.1–53.4 weeks) for patients with no evidence of disease who were transferring surveillance care, less than 0.1 week (range, <0.1–159.0 weeks) for patients who presented with recurrent disease, and 2.4 weeks (range, <0.1–197.9 weeks) for patients who were referred with persistent disease (mean values also shown in Table 2).

Table 2.

Reason for Referral and Time to Diagnosis

Type of Presentation	No. (%) of Patients	Mean/Median Weeks to Diagnosis (Range)
Initial workup	365 (65.3)	11.9/3.9 (<0.1–170.4)
Persistent disease	31 (5.5)	23.8/0.3 (<0.1–53.4)
Recurrent disease	76 (13.4)	10.1/<0.1 (<0.1–197.9)
No evidence of disease	36 (6.4)	6.2/0.3 (<0.1–53.4)
Referred for treatment completion	38 (6.8)	12.6/1.9 (<0.1–185.4)

Presenting with persistent or recurrent disease did not correlate significantly with sex, ethnicity, referring provider type, or subsite of disease (Table 3) when compared with presenting for initial workup and management. Also, whether prereferral management was NCCN compliant did not correlate with any of these factors or with insurance status (Table 3). Furthermore, we examined the types of physicians who were providing prereferral care to those patients who presented with persistent or recurrent disease (Table 4). While the involvement of radiation oncologists or endocrinologists appeared to affect NCCN compliance, neither had a statistically significant impact.

Table 3.

Impact on Disease Presentation and Type of Prereferral Care

Variable	Type of Disease Presentation, No. (%)			P Value	Type of Prereferral Care, No. (%)		P Value
	Initial Presentation	Persistent Disease	Recurrent Disease		NCCN-Compliant Care	NCCN-Noncompliant Care
Sex
Male	228 (71.3)	15 (4.7)	52 (16.3)	.15	36 (56.3)	28 (43.8)	.85
Female	137 (72.9)	16 (8.5)	24 (12.8)		25 (58.1)	18 (42.9)
Ethnicity
White	285 (71.4)	25 (6.3)	62 (15.5)	.88	48 (56.5)	37 (43.5)	.41
Hispanic	38 (74.5)	4 (7.8)	5 (9.8)		6 (60.0)	4 (40.0)
Black	31 (72.1)	2 (4.7)	7 (16.3)		4 (44.4)	5 (55.6)
Asian	10 (71.4)	0	2 (14.3)		3 (100)	0
Referring physician
Otolaryngologist	155 (77.9)	11 (5.5)	28 (14.1)	.05	24 (68.6)	11 (31.4)	.54
Primary care	48 (78.7)	3 (4.9)	6 (9.8)		6 (66.7)	3 (33.3)
Endocrinologist	13 (50.0)	3 (11.5)	9 (34.6)		6 (50.0)	6 (50.0)
Oncologist	10 (50.0)	3 (15.0)	6 (30.0)		4 (40.0)	6 (60.0)
Dermatologist	28 (73.7)	2 (5.3)	6 (15.8)		4 (50.0)	4 (50.0)
General surgeon	3 (75.0)	1 (25.0)	0		1 (50.0)	1 (50.0)
Self-referral	70 (58.3)	8 (6.7)	19 (15.8)		13 (46.4)	15 (53.6)
Sub-site of disease
Oral cavity	51 (76.1)	3 (4.5)	9 (13.4)	.49	6 (50.0)	6 (50.0)	.45
Oropharynx	58 (72.5)	4 (5.0)	7 (8.8)		7 (63.6)	4 (36.4)
Hypopharynx	6 (60.0)	2 (20.0)	2 (20.0)		3 (75.0)	1 (25.0)
Larynx	34 (68.0)	1 (2.0)	11 (22.0)		10 (90.9)	1 (9.1)
Nasopharynx	5 (55.6)	1 (11.1)	3 (33.3)		1 (33.3)	2 (66.7)
Endocrine	52 (65.0)	8 (10.0)	14 (17.5)		13 (56.5)	10 (43.5)
Skin	49 (62.8)	7 (9.0)	18 (23.1)		15 (57.7)	11 (42.3)
Salivary gland	11 (61.1)	1 (5.6)	4 (22.2)		1 (25.0)	3 (75.0)
Paranasal sinuses	5 (1.6)	1 (12.5)	2 (25.0)		3 (75.0)	1 (25.0)
Unknown primary	9 (69.2)	0	3 (23.1)		2 (66.7)	1 (33.3)
Insurance type
Private					37 (64.9)	20 (35.1)	.23
Government					22 (50.0)	22 (50.0)
Self					2 (40.0)	3 (60.0)

Table 4.

Types of Providers Involved in Prereferral Management

	No. (%) of Patients
Type of Physician	NCCN-Compliant Care	NCCN-Noncompliant Care
Otolaryngologist	37 (60.6)	28 (54.3)
Medical oncologist	19 (31.1)	12 (26.1)
Radiation oncologist	29 (47.5)	15 (32.6)
Endocrinologist	10 (16.4)	3 (6.5)
Dermatologist	9 (14.8)	9 (19.6)
General surgeon	4 (6.6)	2 (4.3)
Primary care physician	8 (13.1)	5 (10.9)

Abbreviation: NCCN, National Comprehensive Cancer Network.

Of those patients who presented with recurrent or persistent disease, 43% had prereferral care that was not compliant with NCCN guidelines. Of the 43%, 59% received inadequate surgical treatment, 15% refused appropriate recommendations, and 11% were treated for the wrong diagnosis. Table 5 lists the reasons for discordant care.

Table 5.

Rates of National Comprehensive Cancer Network (NCCN) Noncompliance and Reasons

Variable	No. (%) of Patients
NCCN noncompliance for patients with prereferral treatment	46/107 (43.0)
Reason for noncompliance
Inadequate surgery	27/46 (58.7)
Inadequate adjuvant therapy	5/46 (10.9)
Inadequate radiation therapy	2/46 (4.4)
Misdiagnosis	7/46 (15.2)
Patient refused recommended treatment	5/46 (10.9)

COMMENT

Assessing the quality of care has become a national priority, as espoused by the IOM 2001 report “Crossing the Quality Chasm: A New Health System for the 21st Century.”¹ This report emphasized the importance of improving the effectiveness, timeliness, efficiency, equity of delivery, and safety of medical care to optimize the quality of medical care.^2,3 It also highlighted the need to develop quality indicators to address these issues.

Most studies evaluate the quality of care through examination of mortality, morbidity, and length of hospital stay. However, clinical practice guidelines offer specialty- and disease-specific treatment recommendations that can, in turn, be used to develop criteria for measurable quality.^4,12 To examine the care received by patients with head and neck cancer prior to referral to a tertiary care medical center, we used NCCN guidelines for head and neck cancer management as a quality standard. We found that some aspect of prereferral care was noncompliant with NCCN recommendations in 43% of patients who were referred with persistent or recurrent disease.

Patients who were referred with persistent or recurrent disease accounted for 18.9% (107 of 566) of all new patients who presented to our department over a 4-month period. In attempting to address the issue of timeliness of care, as emphasized by the IOM, we had difficulty accurately assessing the weeks to diagnosis owing to limited available information on prereferral care (Table 2). Although head and neck cancer is relatively rare, cumulatively accounting for fewer than 3% of solid tumors diagnosed annually in the United States, our study data highlight the need for raised awareness among community providers who are evaluating the initial symptoms of these patients.

We did not identify any factors associated with whether prereferral care was noncompliant with NCCN guidelines, despite examining the potential effects of race, sex, insurance status, subsite of disease, or type of referring or treating physicians (Tables 3 and 4). While indicating compliance with the IOM goal of equity of care, our data also suggest that efforts to improve the quality of head and neck cancer care should be globally directed, regardless of physician subspecialty training. It would be expected that subspecialists would compose a greater percentage of referring physicians for patients with persistent or recurrent disease. The lack of a significant difference between referral patterns for patients presenting for initial treatment and those with recurrent or persistent disease indicates that perhaps referral criteria, especially for complicated cases, should be devised. Recently, Cook et al¹⁴ reported that patients with heart disease who were treated at community health centers were significantly less likely to have specialist consultations or follow-up consultations than were those who were treated at an academic medical center. While there are no concrete criteria for specialist referrals, addressing this issue would serve the interest of timeliness of diagnosis and efficiency of care. However, further studies are needed to determine the impact of delayed referral on patient outcomes.

Of the patients who presented with persistent or recurrent disease, 43% had an element of prereferral care that deviated from NCCN guidelines (Table 5). This accounted for 8.1% of all new patient evaluations. Most of these cases (58.7%) were noncompliant by reason of inadequate surgical treatment, as determined by preoperative imaging reports, operative notes, and pathology reports. Misdiagnosis leading to inappropriate treatment occurred in 15.2% of cases. This number underscores the need to raise awareness of head and neck cancer among community providers who are evaluating patients’ presenting complaints. Inadequate primary radiation therapy was identified in 4.4% of cases, and insufficient adjuvant therapy was seen in 10.9%. Surprisingly, 15.2% of noncompliant care was attributable to patient refusal of indicated recommended treatment. This raises questions concerning how informed these patients were in making these decisions and how well they were educated about the ramifications of such decisions.

There are many studies evaluating adherence to established guidelines. The National Initiative for Cancer Care Quality established a set of explicit quality measures for early- and intermediate-stage breast and colorectal cancer, finding that 82% to 87% of patients with breast cancer and 78% of patients with colorectal cancer received care that was compliant with evidence-based guidelines. However, significant variation in adherence was identified, indicating opportunities for improvement.⁹ Patel et al¹² evaluated whether care was concordant with guidelines established by the American College of Cardiology and the American Heart Association for the management of acute coronary syndrome. They found that compliance ranged from 74% to 78%. They also noted that there were slightly higher rates of adherence at academic hospitals and higher rates of variability among nonacademic settings.

While there are many reports of rates of compliance with guidelines, most do not correlate compliance with patient outcome. Kapadia et al⁵ determined a 57% rate of inappropriate carotid endarterectomy, based on recommendations set by the American Academy of Neurology, in an urban community hospital. Furthermore, they found that these nonindicated procedures resulted in a higher rate of complications than indicated cases. Specific to head and neck cancer, Benasso et al¹⁵ reported that patients in a clinical trial of chemoradiotherapy had a significantly better 3-year survival if they were treated by the coordinating medical center instead of by community affiliates. They attributed this to their finding that community hospitals had lower treatment compliance: they treated patients with lower doses of chemotherapy and radiation therapy than the coordinating medical center and were more likely than the coordinating medical center to delay treatment despite no reported differences in toxicity. Furthermore, the community affiliates had much lower volumes of patients (127 patients among 8 treatment centers) as compared with the coordinating medical center (166 patients in 1 treatment center). Interestingly, the authors found no difference among patients treated with radiotherapy alone. Kubicek et al¹⁶ reported no difference in survival or recurrence rates among patients with head and neck cancer who were treated with radiotherapy at academic vs nonacademic settings and found a trend for better survival in patients who received chemoradiotherapy at an academic medical center vs a community setting, although this did not reach statistical significance. Relative to our findings, further study is needed to evaluate how care that deviates from NCCN guidelines affects patient outcomes.

We recognize that there are limitations of this study. Although we evaluated 107 patients, the patient case mix was fairly heterogeneous, with no more than 20% of cases representing a single subsite of disease. While this may pose a limitation in discussing site-specific management, we did not find that disease subsite significantly correlated with whether a patient presented with recurrent or persistent disease or whether a patient was newly diagnosed. Despite our best efforts in systematically contacting both the patient and the referring physician’s office to obtain complete records of prereferral treatment, accessing records was difficult. Evaluations of prereferral care were made based on all available records and information obtained from patient interview. In cases in which such information was not available and care could not be proved to deviate from NCCN guidelines, providers were given the benefit of the doubt and care was presumed to be compliant with NCCN standards. Therefore, our determination of deviation from NCCN recommendations is possibly understated.

The difficulty we faced in obtaining information about prereferral management also emphasizes the need for continuity of information among physicians. We found that complete records, documenting the initial encounter and patient visits through treatment and follow-up, were available for fewer than 5% of patients who were evaluated by a primary care physician and for only 6% of those who were previously treated by an otolaryngologist. However, even in cases in which complete records were not available, often some of the records were obtainable. A recent study evaluated the exchange of information among community providers, finding that physicians had information from previous visits with other practitioners in only 22% of patient visits. The exchange of information was found to be better when it came from consultants to primary care providers rather than vice versa.¹⁷ Continuity of care requires physician continuity, information continuity, and management continuity. A uniform system of electronic medical records would facilitate this flow of information, but until the flow is established, effort must be made by providers to ensure continuity of information to optimize the quality of care that is being delivered to their patients.

We initially attempted to evaluate the percentage of patients who were presented at multidisciplinary conferences but found mention of multidisciplinary conferences to be too sparse to be reliable. Wright et al¹⁸ recently surveyed general surgeons in academic and community practices in Ontario, Canada, and found that 100% of academic and 44% of community hospitals hold multidisciplinary conferences. Importantly, academic clinicians attended these conferences weekly, whereas community providers attended on a monthly basis, even when the conference was held more frequently. Multidisciplinary conferences improve patient care; a study of patients with rectal cancer in a Veterans Affairs hospital found that care was more likely to adhere to National Cancer Institute guidelines if the case was discussed at a multidisciplinary conference.¹⁹ Note must be made that deviation from NCCN guidelines is occasionally intentional, depending on individual patient factors. However, in such cases, the decision to deviate should be reached as a consensus, optimally in a multidisciplinary forum. Presentation of patients with head and neck cancer at multidisciplinary conferences may be a means by which community providers can improve compliance with NCCN guidelines.

Prereferral physicians, however, may not be aware of or practice by the guidelines we used as the standard of quality. Bober et al²⁰ surveyed primary care physicians who were caring for long-term cancer survivors and found that although 82% thought that guidelines for the primary care of cancer survivors were not well defined, more than 90% were not aware of the IOM report describing the health care needs of these patients. They also identified a general lack of awareness about NCCN and American Society of Clinical Onocology survivorship guidelines. We did not survey the physicians who were providing prereferral care to determine their awareness of NCCN guidelines, although our findings suggest that disseminating and encouraging the implementation of these recommendations may improve the care of patients with head and neck cancer.

Our findings support the regionalization of head and neck cancer care. Treatment recommendations from multidisciplinary treatment planning conferences at our tertiary care medical center are in accordance with NCCN or institutional guidelines whenever possible. Also, all eligible patients are offered treatment protocols that have been approved by our institutional review board. Past reports have convincingly shown that medical centers and physicians who are treating a higher volume of patients with a particular disease have a lower mortality, even after adjustment for case mix. These reports have been both regional, encompassing multiple surgical conditions and medical diseases,^21–26 and procedure or disease specific.^27–29 Although the high-volume hospitals in many studies are academic medical centers, the added of benefit of teaching status is not clear.^30,31 It should be noted that volume-outcome relationships reflect a structural characteristic, while NCCN guidelines delineate a process of care. Furthermore, most of these studies implemented short-term measures of morbidity and mortality, which are not appropriately extrapolated to cancer care. However, specific to head and neck cancer care, the previously discussed study by Benasso et al¹⁵ indicates that treatment center volume significantly affects the prognosis of patients with head and neck cancer who are undergoing chemoradiotherapy. To improve the quality of head and neck cancer care through increased compliance with national guidelines, steps need to be taken toward educating community providers through dissemination of guidelines, implementation of multidisciplinary conferences, and regionalization of the care of this patient population.

In conclusion, assessing and optimizing the quality of care has become a national priority. Compliance with clinical practice guidelines, which are built on evidence and expert consensus, can serve as a quality indicator. We found that for patients who were referred to a tertiary care center for persistent or recurrent head and neck cancer, prereferral care deviated from NCCN guidelines in 43% of cases, indicating the need for dissemination and implementation of NCCN recommendations.