Abstract
This study uses National Cancer Database data to estimate associations between hurricane disaster declarations, which could disrupt electrical power, and survival of patients undergoing radiotherapy for nonoperative locally advanced non–small cell lung cancer between 2004 and 2014.
Natural disasters, such as hurricanes, can interrupt the provision of oncology care.1 Radiotherapy is particularly vulnerable because it requires dependable electrical power and daily treatment.2 Disruptions are especially concerning for patients undergoing treatment for locally advanced non–small cell lung cancer (NSCLC)2 because treatment delays as little as 2 days negatively affect survival.3
We investigated whether hurricane disasters occurring during radiotherapy were associated with poorer survival for patients with NSCLC.
Methods
Patients undergoing definitive radiotherapy for nonoperative locally advanced NSCLC between 2004 and 2014 were selected from the hospital-based National Cancer Database, which captures approximately 70% of all cases in the United States and requires hospitals to have 90% annual follow-up of living patients.4 All patients had at least 1 year of follow-up for vital status (up to December 31, 2015). Disaster declarations were identified from the Federal Emergency Management Agency for 2004 to 2014.5
Exposed patients were undergoing radiation treatment when a hurricane disaster was declared for the facility’s area between the date when radiotherapy started and the date when radiotherapy ended. They were propensity score–matched6 to unexposed patients, who completed treatment at the same facility at times when no disaster was declared, on radiotherapy start month, sex, age, stage, tumor spread to lymph nodes, and zip code–level median income quintile. Pearson χ2 or t tests were used to compare groups.
Overall survival was defined as the interval between age at diagnosis and age at death or last contact. Multivariable Cox proportional hazards modeling included an indicator variable for hurricane disaster declared during radiotherapy, sex, race/ethnicity, income, geographic region, health insurance, comorbidities, tumor size, tumor spread to lymph nodes, facility type, driving distance to facility, receipt of concomitant chemotherapy, number of treatment sessions (fractions) received, and radiotherapy start month and year (2004-2009 and 2010-2014). Proportionality assumption, tested using Schoenfeld residuals, was met. Restricted cubic spline regression flexibly modeled the association between the number of days disaster declarations lasted and survival. Statistical significance was set at a 2-sided α = .05. All analyses were performed using SAS version 9.4. This study was granted exempt review by the institutional review board at the Morehouse School of Medicine.
Results
There were 1934 patients who had a hurricane disaster declared during radiation treatment and 129 080 who completed radiation treatment in the absence of a disaster declaration, with 1734 in the exposed group and 1734 in the unexposed group after matching characteristics were balanced (Table). The 101 disaster declarations lasted between 1 and 69 days.
Table. Characteristics of Patients Exposed to a Hurricane Disaster Declaration During Radiation Treatment for Locally Advanced Non–Small Cell Lung Cancer and Propensity-Matched Unexposed Patients.
| Patient Characteristicsa | Exposed (n = 1734) | Unexposed (n = 1734) | P Valueb |
|---|---|---|---|
| Age at diagnosis, mean (SD), y | 66.5 (9.7) | 66.4 (9.9) | .67 |
| Sex, No. (%) | .97 | ||
| Male | 953 (55.0) | 954 (55.0) | |
| Female | 781 (45.0) | 780 (45.0) | |
| Race/ethnicity, No. (%)c | .16 | ||
| Non-Hispanic white | 1355 (78.5) | 1384 (80.0) | |
| Hispanic | 70 (4.1) | 53 (3.1) | |
| Non-Hispanic black | 271 (15.7) | 273 (15.8) | |
| Non-Hispanic other | 30 (1.7) | 19 (1.1) | |
| Median income quintile, No. (%), $ | .05 | ||
| <36 000 | 326 (19.2) | 336 (19.8) | |
| 36 000-43 999 | 342 (20.1) | 324 (19.1) | |
| 44 000-52 999 | 386 (22.7) | 326 (19.2) | |
| 53 000-68 999 | 336 (19.8) | 381 (22.4) | |
| ≥69 000 | 309 (18.2) | 331 (19.5) | |
| Insurance, No. (%) | .43 | ||
| Private | 520 (30.6) | 570 (33.3) | |
| Uninsured | 73 (4.3) | 61 (3.6) | |
| Medicaid | 130 (7.6) | 122 (7.1) | |
| Medicare | 960 (56.4) | 944 (55.1) | |
| Other | 19 (1.1) | 17 (1.0) | |
| Comorbidity, No. (%) | .52 | ||
| 0 | 1117 (64.4) | 1133 (65.3) | |
| 1 | 442 (25.5) | 415 (23.9) | |
| ≥2 | 175 (10.1) | 186 (10.7) | |
| Tumor spread to lymph nodes, No. (%) | 379 (21.9) | 358 (20.6) | .38 |
| Tumor size, mean (SD), mm | 50.8 (46.5) | 48.6 (38.6) | .17 |
| Treatment duration, mean (SD), d | 46.2 (25.6) | 66.8 (77.8) | <.001 |
| Region, No. (%) | .99 | ||
| Northeast | 622 (35.9) | 625 (36.0) | |
| Midwest | 20 (1.2) | 20 (1.2) | |
| South | 1092 (63.0) | 1089 (62.8) | |
| Facility type, No. (%) | >.99 | ||
| National Cancer Institute–designated | 119 (7.3) | 118 (7.2) | |
| Comprehensive | 845 (51.7) | 847 (51.8) | |
| Teaching | 379 (23.2) | 379 (23.2) | |
| Community | 137 (8.4) | 136 (8.3) | |
| Other | 155 (9.5) | 156 (9.5) | |
| Driving distance, mean (SD), miles | 20.4 (42.6) | 27.0 (103.6) | .02 |
| Concomitant chemotherapy, No. (%) | 485 (28.0) | 497 (28.7) | .65 |
| Fractions of treatment, No. (%) | .54 | ||
| 30 | 126 (22.7) | 122 (20.5) | |
| 33 | 145 (26.2) | 164 (27.6) | |
| 34 | 39 (7.0) | 54 (9.1) | |
| 35 | 138 (24.9) | 130 (21.9) | |
| 36 | 27 (4.9) | 34 (5.7) | |
| 37 | 59 (10.6) | 61 (10.3) | |
| 38 | 20 (3.6) | 29 (4.9) | |
| Month radiotherapy started, No. (%) | .22 | ||
| November-April | 34 (2.0) | 30 (1.7) | |
| May | 56 (3.2) | 43 (2.5) | |
| June | 140 (8.1) | 110 (6.3) | |
| July | 437 (25.2) | 428 (24.7) | |
| August | 558 (32.2) | 614 (35.4) | |
| September | 273 (15.7) | 273 (15.7) | |
| October | 236 (13.6) | 236 (13.6) | |
| Year radiotherapy started, No. (%) | .73 | ||
| 2004-2009 | 953 (55.0) | 943 (54.4) | |
| 2010-2014 | 781 (45.0) | 791 (45.6) |
Exposed patients were undergoing radiation treatment for locally advanced non–small cell lung cancer when a hurricane disaster was declared for the facility’s area. Unexposed patients completed treatment at the same facility not during a disaster declaration. Unexposed patients were propensity-matched to exposed patients on month of initiation of radiation treatment, sex, age, lymph node involvement, and income. Patients were excluded if treated at a facility other than the reporting facility.
Differences between exposed and matched unexposed patients were assessed using Pearson χ2 test for categorical variables and t test for continuous variables.
Because race has a significant association with cancer outcomes, race was coded following the Surveillance, Epidemiology, and End Results program’s coding manual, which uses patients’ self-declared identification as the highest-priority source followed by documentation in the medical record and death certificate.
The median observation time was 15 months. For the exposed group, the total number of deaths was 1408, mean survival time was 29 months, and 5-year survival estimate was 14.5%. For the unexposed group, the total number of deaths was 1331, mean survival time was 31 months, and 5-year survival estimate was 15.4%. Patients affected by a hurricane disaster had longer radiation treatment durations (66.9 vs 46.2 days; P < .001) and significantly worse overall survival than matched unexposed patients in both crude (hazard ratio [HR] for death, 1.11 [95% CI, 1.02-1.22]; P = .02) and adjusted (HR, 1.19 [95% CI, 1.07-1.32]; P = .001) analyses. The adjusted relative risk for death increased with the length of the disaster declaration (Figure, A), reaching 1.27 (95% CI, 1.12-1.44) for disasters lasting 27 days. The association became nonsignificant after 30 days, but only 19 declarations lasted that long (Figure, B).
Figure. Association Between Length of Hurricane Disaster Declaration and Risk of Death in Patients With Lung Cancer Undergoing Radiation .
In panel A, cubic spline regression modeled a 1-unit increase in the number of days the declaration lasted and the overall survival, adjusted for sex, race/ethnicity, income, geographic region, health insurance, comorbidities, tumor size, tumor spread to lymph nodes, facility type, driving distance to facility, receipt of concomitant chemotherapy, number of treatment sessions (fractions) received, and radiation treatment start month and year (2004-2009 and 2010-2014). Only the 1734 patients who were affected by a hurricane disaster declared during radiation treatment were included in this analysis. The solid line represents the relative risk and the dotted lines represent 95% CIs.
Discussion
Having a hurricane disaster declared during radiotherapy was associated with worse overall survival in patients with locally advanced NSCLC. Longer declarations were associated with worse survival.
Strengths of this study include a large national sample with detailed sociodemographic, clinical, and treatment information and adequate follow-up periods. Limitations include lack of information about smoking history, performance status, treatment toxicity, reasons for or exact dates of treatment breaks, and other hurricane disaster–associated factors (eg, displacement, mental health status, physical functioning).
Because data on other potentially explanatory factors are lacking, the relative contribution of treatment delay to the observed association cannot be quantified. However, treatment delay is one of the few hurricane-related disruptions that can be prevented. Because no recommended correction for radiotherapy delays exists,3 strategies for identifying patients, arranging for transferring treatment, and eliminating patient out-of-network insurance charges should be considered in disaster mitigation planning. Research is needed to evaluate other types of natural disasters, diseases, and treatments.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.Man RX-G, Lack DA, Wyatt CE, Murray V. The effect of natural disasters on cancer care: a systematic review. Lancet Oncol. 2018;19(9):e482-e499. doi: 10.1016/S1470-2045(18)30412-1 [DOI] [PubMed] [Google Scholar]
- 2.Roach MC, Robinson CG, Bradley JD. Natural disasters and the importance of minimizing subsequent radiation therapy interruptions for locally advanced lung cancer. Int J Radiat Oncol Biol Phys. 2018;100(4):836-837. doi: 10.1016/j.ijrobp.2017.11.001 [DOI] [PubMed] [Google Scholar]
- 3.McMillan MT, Ojerholm E, Verma V, et al. Radiation treatment time and overall survival in locally advanced non-small cell lung cancer. Int J Radiat Oncol Biol Phys. 2017;98(5):1142-1152. doi: 10.1016/j.ijrobp.2017.04.004 [DOI] [PubMed] [Google Scholar]
- 4.Mallin K, Browner A, Palis B, et al. Incident cases captured in the national cancer database compared with those in U.S. population based central cancer registries in 2012-2014. Ann Surg Oncol. 2019;26(6):1604-1612. doi: 10.1245/s10434-019-07213-1 [DOI] [PubMed] [Google Scholar]
- 5.FEMA disaster declarations summary: open government dataset. Federal Emergency Management Agency website. https://www.fema.gov/media-library/assets/documents/28318#. Accessed September 27, 2018.
- 6.Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399-424. doi: 10.1080/00273171.2011.568786 [DOI] [PMC free article] [PubMed] [Google Scholar]