Temporal Trends and Predictors of Perioperative Chemotherapy Use in Elderly Patients With Resected Nonsmall Cell Lung Cancer

Abstract

BACKGROUND

The authors assessed patterns of perioperative chemotherapy use in elderly patients with resected stage I, II, or IIIA nonsmall cell lung cancer (NSCLC) from 1992 to 2002.

METHODS

By using data from the Surveillance, Epidemiology, and End Results Program, 11,807 patients were identified who had resected stage I, II, or IIIA NSCLC between 1992 and 2002 and survived ≥120 days beyond diagnosis. The rate of perioperative chemotherapy use was measured by calendar year, and the association between clinical/demographic characteristics and the receipt of chemotherapy was examined by using logistic regression.

RESULTS

In total, 957 patients with stage I, II, or IIIA NSCLC (8.1% of the study population) received perioperative chemotherapy. The proportion of patients receiving chemotherapy for stage I NSCLC changed little during the study period. Of 3230 patients with stage II and IIIA NSCLC, 609 patients (18.9%) received chemotherapy, 423 patients (13%) received chemotherapy combined with radiation. 452 patients (15.6%) received adjuvant chemotherapy, and 66 patients (2.3%) received neoadjuvant chemotherapy. The use of chemotherapy increased significantly among patients who were diagnosed after 1994 relative to patients who were diagnosed in 1992 after controlling for sociodemographic and treatment characteristics (P <.001). There was significantly increased use of new-generation chemotherapy agents, such as carboplatin and taxanes (P <.001). The proportion of patients receiving combined-modality therapy also increased significant (P <.001). Younger age, being married, having advanced-stage tumor or adenocarcinoma, having a later diagnosis year, receiving radiation, and seeing an oncologist were predictors for the receipt of chemotherapy (P <.001).

CONCLUSIONS

A substantial proportion of Medicare beneficiaries with NSCLC received perioperative chemotherapy. Specifically designed prospective trials that focus on older patients are needed.

Lung cancer is the most common malignancy and the leading cause of cancer-related death in the U.S., with 174,470 new diagnoses and 163,460 deaths in 2006.¹ The majority of patients with lung cancer have nonsmall cell lung cancer (NSCLC).² The most effective treatment for stage I, II, and IIIA NSCLC is surgical resection. Despite surgical resection, from 50% to 60% of patients relapse and die from their lung cancer.³ Systemic recurrence has been reported as the most frequent type of failure in patients with NSCLC,⁴ and it is postulated that >75% of deaths are related to disease relapse, especially distant metastases.⁵ Because of these recurrences, the 5-year survival rate of patients with early-stage NSCLC, despite complete resection, ranges from 45% to 70%.^5,6 After other solid cancers (ie, breast, colon), effective adjuvant therapy theoretically should include systemic chemotherapy. Many attempts have been made in the past to reduce the risk of relapse and death from lung cancer by administering adjuvant chemotherapy to patients after surgical resection.

In 1995, a meta-analysis of postoperative chemotherapy in NSCLC by the British Medical Council indicated a 13% reduction in the hazard of death, leading to a 5% absolute improvement in survival 5 years after the start of adjuvant cisplatin-based chemotherapy compared with observation only.⁷ Although this trend did not reach statistical significance, the findings led to multiple national and international trials that were powered adequately to find an absolute benefit of 5% at 5 years. Results of these trials have been reported in the last 5 years.^8–16

In 2003, the abstract of the International Adjuvant Lung Cancer Trial (IALT) study, which was presented at the Annual Meeting of American Society of Clinical Oncology, first demonstrated a survival benefit of adjuvant chemotherapy.⁸ In 2004, chemotherapy became the standard of care in the U.S., when the IALT trial demonstrated that, compared with surgery alone, cisplatin-based adjuvant therapy improved the 5-year survival rate by 40% to 44.5% in patients with resected NSCLC.⁹

To our knowledge, there is little information regarding the use of adjuvant chemotherapy in elderly patients with resectable NSCLC outside clinical trials. It is known that the elderly are underrepresented in clinical trials.¹⁷ For example, the median age in the IALT trial was 59 years (range, 27–77 years).^8,9 We used data from the population-based Surveillance, Epidemiology, and End Results (SEER)-Medicare database to study the trend in receipt of chemotherapy during the decade from 1992 to 2002 to investigate factors associated with use of perioperative chemotherapy among elderly patients with resected NSCLC.

MATERIALS AND METHODS

Data Source and Cohort

The study cohort was comprised of patients who were registered in the National Cancer Institute’s SEER Program from 1992 to 2002. The SEER Program collects uniformly reported data from 11 population-based cancer registries covering approximately 14% of the U.S. population.¹⁸ For each incident cancer, the SEER registries collect information on patients’ tumor characteristics, demographic characteristics, and month and year of diagnosis. Since 1991, SEER data have been merged with Medicare administrative data by a matching algorithm that has successfully linked files for >94% of SEER registry patients who were diagnosed at age ≥65 years. The Medicare claims files contain extensive diagnostic, treatment, and cost data for patients who are covered by Medicare. In 1993, these 2 databases were linked.¹⁹

We identified 34,511 patients aged >66 years who were diagnosed with NSCLC between 1992 and 2002 (SEER codes 34.0–34.9 and International Classification of Diseases for Oncology, Second edition [ICD-O-2] morphology codes 8010–8040, 8050–8076, 8140, 8250–8260, 8310, 8320, 8323, 8430, 8470–8490, 8550–8573, 8980, and 8981). We excluded patients without Medicare Part A and Part B coverage during the year before and the 120 days after diagnosis and those who were members of a health maintenance organization (n = 9709 patients for both exclusions).

Identification of Staging and Surgical Procedures

We defined resection surgery as wedge resection, pneumonectomy, or lobectomy that occurred within 4 months after the diagnosis of NSCLC from both the site-specific surgery code in SEER and Medicare claims. The IDC ninth revision, clinical modification (ICD-9-CM) and Current Procedural Terminology (CPT) codes were taken from outpatient and inpatient billing claims and were used to define the following procedures: pneumonectomy (ICD9-CM codes 32.50 and 32.60; CPT codes 32,440, 32,442, and 32,445), lobectomy (ICD9-CM code 32.40; CPT codes 32,480, 32,482, 32,484, and 32,486), and wedge resection (ICD9-CM codes 32.29 and 32.30; CPT code 32,500). Of 24,802 patients who met the selection criteria, 11,807 patients underwent resection.

Identification of Chemotherapy and Radiation Treatment

We used the Health Care Financing Administration’s Common Procedure Coding System (HCPCS) to identify patients who received a specific chemotherapeutic drug within 120 days after their NSCLC diagnosis according to their physician claims or their hospital outpatient claims files. We chose the 120-day window to differentiate initial adjuvant treatment from treatment at relapse. In addition, using Level II HCPCS codes (J9XXX, 96,400–96,549), ICD-9-CM diagnostic codes (V581, V662, and V672), revenue codes (0331, 0332, and 0335), and a procedure code (9925) from physician claims files or outpatient or admission hospital claims, we searched for evidence of other chemotherapy delivery. Among patients without specific J9XXX chemotherapy HCPCS codes, these codes were used to capture evidence that an unspecified chemotherapeutic drug had been given during the 120 days after diagnosis. We defined radiation therapy within 120 days after cancer diagnosis through the radiation code in SEER and Medicare claims.

Identification of Patients Seeing a Medical Oncologist

We defined the patient as one who “saw a medical oncologist” if the patient had a physician claim within the 4-month period from the month of diagnosis and the physician specialty (primary or secondary) was medical oncology or hematology based on data from the Centers for Medicare and Medicaid Services (CMS). The physician claims have a 2-digit CMS provider specialty code (90 for medical oncologist, 83 for hematology oncologist) that represents the specialty reported to the carrier who processed the claim.¹⁸

Definition of Explanatory Variables

Our analysis of use of adjuvant chemotherapy by year was adjusted by the following patient-level variables: sex, race, marital status, age at diagnosis, American Joint Committee on Cancer stage, tumor grade, histology type, and SEER site. The socioeconomic characteristics of each patient were based on the percentage of adults with <12 years of education and the percentage of residents living below the poverty level from census tract data. To assess the prevalence of comorbid disease in our cohort, we calculated the Charlson comorbidity index by identifying billing codes for various conditions during the year before diagnosis of cancer using the Deyo implementation of the Charlson score applied to both inpatient and outpatient claims.¹⁹ Hospital affiliation with a medical school was available from the SEER-Medicare Hospital file and was categorized as no, limited, graduate, and major affiliation. The number of beds at each hospital also was available from the SEER-Medicare Hospital file.

Statistical Analysis

Statistical analyses were performed using SAS software (version 9.13; SAS Inc., Cary, NC). We measured the proportion of patients with resected NSCLC who underwent adjuvant chemotherapy in each calendar year stratified by stage. We performed univariate and multivariate logistic regression analyses to evaluate the association between other clinical and demographic characteristics and adjuvant chemotherapy use. Covariates included sex, race, age, marital status, SEER registry, median household income, tumor stage at diagnosis, tumor grade, and histology.

RESULTS

Receipt and Predictors of Perioperative Chemotherapy

In total, 11,807 patients with stage I, II, or IIIA underwent surgical resection during the study period. Among them, 957 patients (8.1% of the study population) received perioperative chemotherapy. The proportion of patients with stage I NSCLC receiving chemotherapy changed little over the study period (Fig. 1). Next, we focused our attention on the 3230 patients with stage II or IIIA NSCLC who underwent resection. Table 1 shows the demographic and clinical characteristics of those 3230 patients and the percentage in each subgroup who received chemotherapy. Overall, 609 patients (18.9%) received perioperative chemotherapy. Chemotherapy was delivered postoperatively to 15.6% of patients and preoperatively to 2.3% of patients. Combined-modality therapy (chemotherapy and radiation therapy) were delivered to 423 patients (13%), and chemotherapy alone was delivered to 186 patients (5.8%).

FIGURE 1.

Time trends in Medicare claims for perioperative chemotherapy by stage during the period from 1992 to 2002. AJCC indicates American Joint Committee on Cancer.

TABLE 1.

Demographic and Clinical Characteristics of Patients With Resected Stage II, IIIA Nonsmall Cell Lung Cancer and the Percentage Treated With Adjuvant Chemotherapy During the Period 1992–2002

Characteristic	Total no. of patients	Adjuvant chemotherapy		P
		No. of patients	%
Total	3230	609	18.9
Age, y
66–69	938	242	25.8	<.0001
70–74	1148	227	19.8
75–79	787	108	13.7
≥80	357	32	9
Sex
Men	1804	349	19.3	.4219
Women	1426	260	18.2
Race
White	2793	534	19.1	.4837
Black	164	33	20.1
Hispanic	99	15	15.2
Other/unknown	174	27	15.5
AJCC stage
II	1624	216	13.3	<.0001
IIIA	1606	393	24.5
Grade
Well differentiated	169	34	20.1	.1625
Moderate differentiated	953	167	17.5
Poorly differentiated	1788	334	18.7
Unknown	320	74	23.1
Histologic type
Squamous cell carcinoma	1025	157	15.3	.0003
Adenocarcinoma	1257	270	21.5
Bronchioalveolar	206	29	14.1
Others	742	153	20.6
SEER region
San Francisco	224	30	13.4	.0011
Connecticut	510	83	16.3
Michigan	574	139	24.2
Hawaii	86	13	15.1
Iowa	398	62	15.6
New Mexico	101	14	13.9
Seattle	362	73	20.2
Utah	56	7	12.5
Georgia	206	50	24.3
San Jose	161	35	21.7
Los Angeles	552	103	18.7
Census tract education (% of adults with<12 y of education)
<11	694	141	20.3	.2164
11–18	696	137	19.7
18–26	673	124	18.4
≥26	696	113	16.2
Census tract poverty level (% living below the poverty line)
<3.5	728	155	21.3	.1170
3.5–6.5	708	117	16.5
6.5–12	666	118	17.7
≥12	657	125	19
Comorbidity index
0	2158	429	19.9	.1945
1	719	123	17.1
2	242	38	15.7
≥3	111	19	17.1
Married
No	1223	194	15.9	.0007
Yes	2007	415	20.7
Metropolitan area
No	396	75	18.9	.9632
Yes	2834	534	18.8
Teaching hospital*
Major	1127	181	16.1	.0362
Limited/graduate	673	106	15.8
No affiliation	1098	216	19.7
No. of hospital beds*
<350	1231	226	18.4	.2209
≥350	1667	277	16.6
Seeing oncologists
Yes	1540	485	31.5	<.0001
No	1690	124	7.3

AJCC indicates American Joint Committee on Cancer; SEER, Surveillance, Epidemiology, and End Results Program.

^*For 9% of patients, surgical procedures could not be found in inpatient claims from 1 month before to 4 months after lung cancer diagnosis.

Younger patients were more likely than older patients to receive chemotherapy. For patients ages 66 to 69 years and aged ≥80 years, chemotherapy use declined from 25.8% to 9% (P <.0001). Among patients who had stage II NSCLC, 13.3% received chemotherapy compared with 24.5% of patients who had stage IIIA NSCLC (P <.0001). In addition to age and advanced stage, having an adenocarcinoma histology (P <.0001), having a later diagnosis year (P <.0001), being married (P <.0001), receiving radiation (P <.0001), and seeing an oncologist (P <.0001) were associated with chemotherapy use in univariate analysis (Table 1). Multivariate analyses generally confirmed the results of the univariate analyses (Table 2).

TABLE 2.

Multivariate Logistic Regression Analysis of the Factors Associated With the Receipt of Chemotherapy in Patients With Stage II, IIIA Nonsmall Cell Lung Cancer

Characteristic	Receiving adjuvant chemotherapy: OR (95% CI)
	Model I	Model II
Age
Every 5 y	0.67 (0.60–0.75)	0.68 (0.61–0.77)
Sex
Men	1.00	1.00
Women	0.98 (0.80–1.21)	0.95 (0.76–1.18)
Race
White	1.00	1.00
Black	0.87 (0.55–1.37)	1.00 (0.62–1.62)
Hispanic	0.61 (0.33–1.12)	0.58 (0.30–1.09)
Other/unknown	0.69 (0.41–1.17)	0.72 (0.41–1.26)
AJCC stage
II	1.00	1.00
IIIA	1.79 (1.47–2.18)	1.74 (1.42–2.14)
Grade
Well differentiated	1.00	1.00
Moderate differentiated	0.78 (0.50–1.22)	0.91 (0.57–1.46)
Poorly differentiated	0.78 (0.51–1.20)	0.88 (0.56–1.38)
Unknown	0.99 (0.60–1.63)	1.09 (0.64–1.85)
Histologic type
Squamous cell carcinoma	1.00	1.00
Adenocarcinoma	1.57 (1.24–1.99)	1.42 (1.11–1.82)
Bronchioalveolar	0.98 (0.61–1.57)	0.90 (0.55–1.47)
Others	1.40 (1.07–1.84)	1.39 (1.04–1.85)
SEER region
Utah	1.00	1.00
San Francisco	1.47 (0.58–3.73)	2.03 (0.77–5.32)
Connecticut	1.70 (0.71–4.10)	2.31 (0.93–5.71)
Michigan	2.78 (1.17–6.57)	2.95 (1.22–7.15)
Hawaii	1.64 (0.52–5.14)	1.46 (0.44–4.78)
Iowa	1.68 (0.70–4.04)	1.61 (0.65–3.96)
New Mexico	1.44 (0.52–4.01)	1.44 (0.50–4.15)
Seattle	2.47 (1.03–5.95)	2.87 (1.16–7.07)
Georgia	2.88 (1.17–7.12)	2.28 (0.90–5.78)
San Jose	2.83 (1.12–7.15)	3.99 (1.53–10.43)
Los Angeles	2.59 (1.09–6.15)	2.37 (0.98–5.77)
Year of diagnosis
1992	1.00	1.00
1993	1.43 (0.85–2.14)	1.29 (0.74–2.23)
1994	1.89 (1.14–3.12)	1.84 (1.08–3.11)
1995	2.57 (1.57–4.21)	1.90 (1.13–3.19)
1996	2.84 (1.75–4.63)	2.14 (1.28–3.57)
1997	2.44 (1.48–4.03)	1.77 (1.05–2.99)
1998	4.10 (2.50–6.72)	2.87 (1.71–4.84)
1999	4.43 (2.66–7.38)	3.40 (1.98–5.82)
2000	4.72 (2.85–7.83)	3.57 (2.09–6.09)
2001	3.44 (1.85–6.38)	2.30 (1.20–4.40)
2002	4.22 (2.11–8.43)	2.77 (1.33–5.76)
Comorbidity index
0	1.00	1.00
1	0.85 (0.67–1.08)	0.84 (0.65–1.07)
2	0.77 (0.52–1.14)	0.73 (0.49–1.10)
≥3	0.85 (0.50–1.45)	0.81 (0.46–1.42)
Married
No	1.00	1.00
Yes	1.32 (1.07–1.64)	1.29 (1.03–1.61)
Census tract poverty level (% living below the poverty line)
<3.5	1.00	1.00
3.5–6.5	0.83 (0.62–1.11)	0.76 (0.56–1.04)
6.5–12	0.88 (0.65–1.19)	0.81 (0.59–1.12)
≥12	1.02 (0.74–1.42)	1.06 (0.75–1.49)
Teaching hospital*
Major	1.00	1.00
Limited/graduate	1.02 (0.76–1.37)	1.06 (0.78–1.45)
No affiliation	1.20 (0.90–1.60)	1.12 (0.83–1.51)
No. of hospital beds*
<350	1.00	1.00
≥350	0.92 (0.72–1.17)	0.88 (0.68–1.13)
Radiation
No	1.00	1.00
Yes	2.43 (1.98–2.98)	2.15 (1.74–2.66)
Seeing oncologists
No		1.00
Yes		5.63 (4.46–7.10)

OR indicates odds ratio; 95% CI, 95% confidence interval; AJCC, American Joint Committee on Cancer; SEER, Surveillance, Epidemiology, and End Results Program.

^*For 9% of patients, surgical procedures could not be found in inpatient claims from 1 month before to 4 months after lung cancer diagnosis.

In univariate analysis, certain geographic locations also were associated with a higher rate of adjuvant chemotherapy use, from a high of 24.2% in Michigan to a low of 12.5% in Utah (Table 1). In multivariate analyses, this variable became less significant (Table 2).

Receiving care in a nonteaching hospital also was associated with a higher rate of adjuvant chemotherapy use than receiving care in a teaching hospital in univariate analysis (Table 1). However, this factor became nonsignificant in multivariate analyses (Table 2). Sex, race, comorbidity, tumor grade, urban residence, income, education level, and hospital volume were not associated with use of perioperative chemotherapy after adjusting for other variables in this cohort of elderly patient with resected NSCLC (Table 2).

Trend and Frequency of Perioperative Chemotherapy Use

The proportion of patients who received perioperative chemotherapy and combination therapy changed significantly during the study period. The use of chemotherapy increased significantly in patients who were diagnosed in 1994 (odds ratio [OR], 1.89; 95% confidence interval [95% CI], 1.41–3.12), 1995 (OR, 2.57; 95% CI, 1.57–4.21), 1996 (OR, 2.84; 95% CI, 1.75–4.63), 1997 (OR, 2.44; 95% CI, 1.48–4.03), 1998 (OR, 4.10; 95% CI, 2.50–6.72), 1999 (OR, 4.43; 95% CI, 2.66–7.38), 2000 (OR, 4.72; 95% CI, 2.85–7.83), 2001 (OR, 3.44; 95% CI, 1.85–6.38), and 2002 (OR, 4.22; 95% CI, 2.11–8.43) relative to the proportion of patients who were diagnosed in 1992 after controlling for sociodemographic and treatment characteristics (Table 2). Chemotherapy use initially increased in the early 1990s, reached to its peak around 1997/ 1998 for stage II and around 1999/2000 for stage IIIA, and slowly declined afterward. The use of chemotherapy in patients with stage IIIA NSCLC increased again around 2000 (Fig. 1). The time trend for the proportion of patients who received combined-modality therapy and chemotherapy alone is shown in Figure 2.

FIGURE 2.

Time trends in Medicare claims for perioperative chemotherapy or combined-modality therapy during the period from 1992 to 2002.

Type of Chemotherapeutic Agents Used

The proportion of patients receiving specific chemotherapeutic agents also changed significantly during the study period (Fig. 3), rising from 16.7% to 75% and from 0% to 65.4% for carboplatin and taxane, respectively, and declining from 45% to 3.85% and from 30% to 7% for etoposide and cisplatin, respectively.

FIGURE 3.

Time trends in Medicare claims for specific chemotherapeutic agents for resected stage II, IIIA nonsmall cell lung cancer during the period from 1992 to 2002.

DISCUSSION

The decision to use chemotherapy is complex in the face of uncertain benefit. Surveys reveal that individual physicians and types of specialists differ widely in their beliefs regarding chemotherapy.^20,21 Because of the high risk of recurrence even in patients with resected NSCLC, clinicians are motivated strongly to add systemic therapy to surgery either preoperatively or postoperatively even in patients with early-stage disease.^22,23 Some experts recommended adjuvant chemotherapy for patients with resected NSCLC in the absence of definitive data because of the great risk of distant recurrence.^22,23 Some patients prefer aggressive treatment because of the perceived high risk of recurrence.²⁴

In a population-based sample of Medicare beneficiaries with resected stage I, II, and IIIA NSCLC between 1992 and 2002, we observed that 8.1% of patients received perioperative chemotherapy. The use of chemotherapy increased significantly in patients who were diagnosed after 1994 relative to patients who were diagnosed in 1992 after controlling for sociodemographic and treatment characteristics. Our findings are consistent with recent studies of patterns of care in patients NSCLC conducted in the U.S. and other countries.^25–27 A noteworthy observation from our study was that chemotherapy use initially increased and reached to its peak around 1997/1998 for stage II, around 1999/2000 for stage IIIA, and slowly declined afterward. Increased chemotherapy use in the 1990s corresponded temporally with publication of the British Medical Council’s meta-analysis, which demonstrated the moderate efficacy of adjuvant chemotherapy in patients with resected NSCLC.⁷ The later decline in the use of chemotherapy corresponded to the publication of several randomized trials, which failed to demonstrate the benefit of adjuvant chemotherapy,^16,28 suggesting that community practice is sensitive to new findings from clinical trials. However, some of those trials were criticized for being underpowered for detecting small benefits of adjuvant chemotherapy in this patient population.²⁹

Chemotherapy use declined dramatically with increasing age at diagnosis, consistent with previous findings.^30,31 This may be because of physicians’ perception of decreased chemotherapy tolerance or increased risks of toxicity in elderly patients.³⁰ Older patients with lung cancer often are compromised by the surgery itself and have a prolonged recovery time compared with other cancer operations, such as mastectomy and colectomy. This may interfere with the receipt of chemotherapy. Elderly patients also may be reluctant to trade quality of life for a perceived survival benefit³¹ and may be more likely to refuse chemotherapy.^32,33

We observed significant associations between disease stage and the receipt of chemotherapy. For patients with stage I disease, perioperative chemotherapy use was rare. This is in line with a recent updated analysis of the Cancer and Leukemia Group B trial, which failed to show a significant improvement in overall survival in patients with stage IB NSCLC.¹¹ For patients with stage II or IIIA disease, chemotherapy use was greater and increased substantially by stage. This is consistent with the results from clinical trials in which increasing benefit was reported with advanced stage.^5,6

We observed that married patients received chemotherapy more often, consistent with findings in other cancers.^34,35 The reasons underlying treatment differences by marital status are unclear but may include encouragement or support from patients’ spouses in seeking aggressive treatment³⁶ or the perception of a higher “value” of life in those patients with a spouse and dependents.³⁶ Clinicians also may recommend aggressive therapies more often to married patients.

Patients who had a medical oncology referral were more likely to receive adjuvant chemotherapy (adjusted OR, 5.63; 95% CI, 4.46–7.10). This is in agreement with other studies, which demonstrated that the decision to get a medical oncology referral is a key step in the eventual receipt of chemotherapy.^37–39

From 1992 to 2002, the combination of chemotherapeutic agents also changed. Our results suggest that carboplatin and paclitaxel came into wide use as agents for patients with NSCLC in the U.S. This shift occurred despite lack of definitive data on the efficacy of these agents in the adjuvant setting. Tolerance and toxicity of chemotherapy is a major concern for elderly patient with lung cancer.⁴⁰ Many elderly patients with NSCLC have difficulty tolerating cisplatin-based treatment because of its nephrotoxicity, ototoxicity, and neurotoxicity; cisplatin administration requires aggressive hydration that may be contraindicated in elderly patients because of their comorbidities. Furthermore, cisplatin is also one of the most emetogenetic drugs. The substitution of carboplatin in elderly patients is appealing.²² Although carboplatin does not possess activity equivalent to that of cisplatin in all platinum-sensitive tumors,^41–43 carboplatin still may be an alternative to cisplatin in elderly patients who are poor candidates for cisplatin. With both activity and tolerability demonstrated in the treatment of advanced NSCLC,^44,45 the taxanes (paclitaxel and docetaxel) also became a popular option in the U.S.; with the increased use of newer chemotherapeutic agents, the use of older agents (cyclophosphamide, etoposide, and cisplatin) dropped to a low rate.

The best sequence of surgery and chemotherapy remains controversial. The majority of patients received chemotherapy postoperatively during the study period. The advantage of adjuvant chemotherapy includes no delay of tumor resection and precise pathologic staging. In addition, chemotherapy generally is more effective in treating minimal-volume disease compared with grossly apparent disease.²⁹ The advantages of using neoadjuvant chemotherapy include earlier commencement of systemic therapy, allowing for the surgical resection of tumor that may not have been suitable for surgery at diagnosis. The potential disadvantages of neoadjuvant therapy include delaying potentially curative surgery and increasing postoperative morbidity and mortality.²⁹

The patients in the current analysis were aged ≥66 years. To our knowledge, there have been no trials specifically assessing chemotherapy for NSCLC in older patients. In a subgroup analysis of the National Cancer Institute of Canada Clinical Trials Group study JBR.10, adjuvant vinorelbine and cisplatin improved survival in patients aged >65 years with acceptable toxicity, even though elderly patients received less chemotherapy.⁴⁶ The results from a meta-analysis of NSCLC⁷ indicated that the benefit of adjuvant chemotherapy was independent of age. We examined the association of chemotherapy with survival from lung cancer in the SEER-Medicare data.

There are major challenges to assessing treatment outcomes in observational data, most of which relate to selection biases. For example, a selection bias may be expected for patients with better underlying health to be more likely to receive chemotherapy.⁴⁷ In our analyses, we used multivariable survival analyses and also controlled for propensity to treat. We also stratified the analyses into deaths from cancer versus deaths from other causes. In our analyses, patients with lung cancer who received chemotherapy experienced an approximate 33% lower mortality from noncancer causes. This indicates that there were strong residual selection biases that were not eliminated after controlling for potential confounders and propensity to receive chemotherapy. Consequently, the survival analyses lack validity, and we did not present them.

The current study has several strengths. First, the nationwide Medicare claims data offer a unique opportunity to study the use of adjuvant chemotherapy in elderly patients with NSCLC. Compared with clinical trials data, which are known to underrepresent elderly patients, population studies are more likely to represent all elderly patients in the real world. Second, this large, population-based study conducted in community-dwelling elderly patients accounted for 14% of the U.S. population. The sample size was large enough to allow us to draw conclusions regarding patterns of care in a diverse population and to examine several potentially confounding variables.

Our study had some limitations. First, the study findings may be applicable only to patients who were diagnosed with NSCLC at age ≥66 years who were not health maintenance organization members and had both Medicare Part A and Part B coverage. Chemotherapy use most likely was greater in patients aged <65 years. Second, administrative databases have inherent problems with data collection and coding. Another limitation is that we were unable to address the role of patients’ performance status in perioperative chemotherapy use. Although the diagnoses included in Medicare claims data do allow for an assessment of comorbidity, previous studies indicate that comorbidity may not be associated with performance status.^48,49

Our results demonstrated a trend toward the increased use of chemotherapy and combined-modality therapy in elderly patients with resected NSCLC during the decade from 1992 to 2002. The most frequent use of chemotherapeutic agents also changed. Our analysis assessed practice patterns that existed before the publication of a large clinical trial that confirmed the survival benefits of adjuvant chemotherapy. Consequently, the proportion of patients receiving chemotherapy today may be higher. Better understanding of the biology of lung cancer, the introduction of less toxic regimens, and better supportive care are likely to impact treatment patterns further in elderly patients with NSCLC. Our findings may be used as a baseline against which the benefit of new therapies can be compared.