Association of First Primary Cancer With Risk of Subsequent Primary Cancer Among Survivors of Adult-Onset Cancers in the United States

Key Points

Question  What are the risks of subsequent primary cancers (SPCs) among adult-onset cancer survivors in the United States?

Findings  In this retrospective cohort study that included 1.54 million survivors with a first primary cancer (FPC) between 1992-2011 and who survived at least 5 years, the risk of developing and dying from SPCs was greater than the risk expected in the general population for 18 and 27 of the 30 FPCs among men, respectively, and for 21 and 28 of the 31 FPCs among women, respectively.

Meaning  The findings emphasize the importance of ongoing surveillance and efforts to prevent new cancers among survivors.

Importance  The number of cancer survivors who develop new cancers is projected to increase, but comprehensive data on the risk of subsequent primary cancers (SPCs) among survivors of adult-onset cancers are limited.

Objective  To quantify the overall and cancer type-specific risks of SPCs among adult-onset cancer survivors by first primary cancer (FPC) types and sex.

Design, Setting, and Participants  A retrospective cohort study from 12 Surveillance, Epidemiology, and End Results registries in the United States, that included 1 537 101 persons aged 20 to 84 years diagnosed with FPCs from 1992-2011 (followed up until December 31, 2017) and who survived at least 5 years.

Exposures  First primary cancer.

Main Outcomes and Measures  Incidence and mortality of SPCs per 10 000 person-years; standardized incidence ratio (SIR) and standardized mortality ratio (SMR) compared with those expected in the general population.

Results  Among 1 537 101 survivors (mean age, 60.4 years; 48.8% women), 156 442 SPC cases and 88 818 SPC deaths occurred during 11 197 890 person-years of follow-up (mean, 7.3 years). Among men, the overall risk of developing any SPCs was statistically significantly higher for 18 of the 30 FPC types, and risk of dying from any SPCs was statistically significantly higher for 27 of 30 FPC types as compared with risks in the general population. Among women, the overall risk of developing any SPCs was statistically significantly higher for 21 of the 31 FPC types, and risk of dying from any SPCs was statistically significantly higher for 28 of 31 FPC types as compared with risks in the general population. The highest overall SIR and SMR were estimated among survivors of laryngeal cancer (SIR, 1.75 [95% CI, 1.68-1.83]; incidence, 373 per 10 000 person-years) and gallbladder cancer (SMR, 3.82 [95% CI, 3.31-4.39]; mortality, 341 per 10 000 person-years) among men, and among survivors of laryngeal cancer (SIR, 2.48 [95% CI, 2.27-2.72]; incidence, 336 per 10 000 person-years; SMR, 4.56 [95% CI, 4.11-5.06]; mortality, 268 per 10 000 person-years) among women. Substantial variation existed in the associations of specific types of FPCs with specific types of SPC risk; however, only a few smoking- or obesity-associated SPCs, such as lung, urinary bladder, oral cavity/pharynx, colorectal, pancreatic, uterine corpus, and liver cancers constituted considerable proportions of the total incidence and mortality, with lung cancer alone accounting for 31% to 33% of mortality from all SPCs.

Conclusions and Relevance  Among survivors of adult-onset cancers in the United States, several types of primary cancer were significantly associated with greater risk of developing and dying from an SPC, compared with the general population. Cancers associated with smoking or obesity comprised substantial proportions of overall SPC incidence and mortality among all survivors and highlight the importance of ongoing surveillance and efforts to prevent new cancers among survivors.

Quiz Ref IDThere were approximately 16.9 million cancer survivors in 2019 in the United States.¹ This number is projected to reach 22.1 million in 2030, due to the aging and growth of the population.² Long-term survivors face physical, psychosocial, medical, behavioral, and socioeconomic consequences of cancer and its treatment.³ One medical consequence is an increased likelihood of subsequent diagnosis with another cancer.⁴ In the United States, new malignancies among cancer survivors (ie, subsequent primary cancers [SPCs]) between 2009 and 2013 comprised a substantial proportion of total incident cancers, ranging from 11% in younger adults (aged 20-64 years) to 25% in adults aged 65 years and older.⁵

The National Cancer Institute monograph published in 2006 described the risk of developing SPCs among cancer survivors diagnosed between 1973 and 2000 in 9 Surveillance, Epidemiology, and End Results Program (SEER) registries.⁴ Since then, prevalence of cancer risk factors, patterns of treatment(s) and survivorship care, and survival have changed considerably. Except for studies of survivors of childhood or young adult cancers,^6,7 few large-scale studies comprehensively examined the risk of developing and dying from SPCs among survivors of adult-onset cancers, with recent studies estimating only incidences for a single primary tumor or for a limited number of cancers.^8-10 Furthermore, in 1992, the SEER program expanded its geographic coverage by adding registries for greater geographic and racial diversity and greater generalizability to the United States.

The objective of this study was to conduct a comprehensive and systematic analysis of incidence and mortality data (1992-2017) in the United States to describe the risk patterns of SPCs among long-term survivors (≥5 years postdiagnosis) of adult-onset cancers.

Institutional review board approval and the need for informed consent were exempted by the institutional review board at Morehouse School of Medicine because data were deidentified and publicly available.

Cases of first primary malignant cancers diagnosed among persons aged 20 to 84 years from 1992 until 2011 and recorded in 12 registries and from 1975 until 2011 in 9 registries (eTable 1 in Supplement 1) were classified according to the International Classification of Diseases for Oncology, Third Edition (30 types in men and 31 types in women; eTable 2 in Supplement 1).^11-13 Race was abstracted from medical records or death certificates and classified as 5 groups (White, Black, American Indian/Alaska Native, Asian/Pacific Islander, unknown) (eMethods in Supplement 1). This information was used to control for differences in the racial distribution between survivors and the general population. The Alaska registry was excluded because of incompatibility of the reference file in the race matching.

SPCs were ascertained following the SEER rules for determining multiple primaries.¹⁴ The category of overall SPCs included all new primary cancers except same-type cancer (to minimize bias due to misclassification of recurrence) and nonmelanoma skin cancer. Follow-up of survivors began 5 years after their first primary cancer (FPC) diagnosis and continued until death, loss to follow-up, attaining age 89 (due to concerns regarding underascertainment in the elderly population),⁴ or December 31, 2017, whichever came first. Multiple SPC diagnoses in a survivor were included to consider tumors that develop after long latency such as those induced by radiation or occurring more commonly at older ages.¹⁵

Sensitivity analyses assessed the overall risks of developing SPCs when restricting to only the first SPC (disallowing multiple events in an individual) or by starting follow-up at 1 year after the FPC diagnosis.

The primary outcomes were incidence (per 10 000 person-years) and relative risk of developing SPCs among 5-year survivors of FPCs (standardized incidence ratio [SIR]). The secondary outcomes were mortality and relative risk of dying from an SPC among survivors (standardized mortality ratios [SMR]).

Incidence rates of SPCs among each group of survivors or survivors overall were obtained by dividing the observed number of SPCs by the corresponding total person-years of follow-up and then multiplying by 10 000. The percentage contribution of any given FPC and SPC combination to the total SPC incidence was calculated among all survivors. The SIR (95% CI) was calculated as the ratio of the observed to the expected number of SPCs.⁴ The expected numbers of SPCs were computed by a weighted sum of stratified incidence rates by sex, age, race, and calendar year from the reference population, which may include multiple primary cancers in an individual. The corresponding excess risks in absolute terms were also calculated and presented in eMethods and eFigure 1 in Supplement 1. All estimates were calculated for overall SPCs and for type-specific (organ-specific) SPCs.⁴ Due to the substantially lower than expected risk for SPCs after prostate cancer, the estimates for male survivors were also presented after excluding prostate cancer survivors. The mortality and SMR of SPCs were calculated using a similar approach.

Among survivors of the 12 smoking-related cancers, per a report from the US Surgeon General,¹⁶ and 12 obesity-related cancers designated by the International Agency for Research on Cancer (eTable 3 in Supplement 1),¹⁷ incidences and SIRs were estimated to quantify the risk of developing any types of smoking-related and obesity-related SPCs. Similar analyses were also performed for 7 alcohol-related¹⁸ and 10 infection-related¹⁹ cancers (eMethods in Supplement 1).

Multicovariate Poisson regression was used to examine SIR trends by diagnosis years (1975-1989, 1990-1999, 2000-2011) for 6 SPCs after the 16 most common FPC types (eMethods in Supplement 1).²⁰

All analyses were stratified by sex and performed using the “MP-SIR” session of SEER*Stat 8.3.6,²¹ SAS 9.4, or Microsoft Excel Office 365 Proplus. All statistical tests were 2-sided, and P values of less than .05 were considered statistically significant, without multiple comparison corrections as the analyses were exploratory.

Among 1 537 101 survivors (mean age, 60.4 years; 48.8% women; 81.5% White), 41% were aged 65 years or older at FPC diagnosis, and 58.8% lived longer than 10 years after FPC diagnosis, accruing 11 197 890 person-years of follow-up (mean, 7.3 years) (Table 1; eTable 4 in Supplement 1). Among male survivors (excluding those with prostate cancer), there were 49 065 SPC cases (177 per 10 000 person-years) and 28 463 SPC deaths (104 per 10 000 person-years) (Table 2), which corresponded with an 11% higher risk of developing SPCs (SIR, 1.11 [95% CI, 1.10-1.12]) and a 45% higher risk of dying from SPCs (SMR, 1.45 [95% CI, 1.43-1.46]) among survivors compared with the risk in the general population. Among women, there were 62 348 SPC cases (109 per 10 000 person-years) and 34 879 SPC deaths (62 per 10 000 person-years), which corresponded with a 10% higher risk of developing SPCs (SIR, 1.10 [95% CI, 1.09-1.11]) and a 33% higher risk of dying from SPCs (SMR, 1.33 [95% CI, 1.32-1.34]).

Most survivor groups defined by FPC types and sex had higher than expected risks of developing or dying from SPCs as compared with the risks in the general population. Among men, the risk of developing any SPCs was statistically significantly higher for 18 of the 30 FPC types, and risk of dying from any SPCs was statistically significantly higher for 27 of the 30 FPC types. Among women, the risk of developing any SPCs was statistically significantly higher for 21 of the 31 FPC types, and risk of dying from any SPCs was statistically significantly higher for 28 of the 31 FPC types (Table 2; eTable 5 in Supplement 1). Among FPC types, the greatest SIR in men was observed after laryngeal cancer (SIR, 1.75 [95% CI, 1.68-1.83]; incidence, 373 per 10 000 person-years), followed by survivors of Hodgkin lymphoma (SIR, 1.59 [95% CI, 1.47-1.72]; incidence, 101 per 10 000 person-years). Among FPC types, the greatest SIR in women was observed after laryngeal cancer (SIR, 2.48 [95% CI, 2.27-2.72]; incidence, 336 per 10 000 person-years) followed by survivors of esophageal cancer (SIR, 1.89 [95% CI, 1.58-2.25]; incidence, 275 per 10 000 person-years). The greatest SMR among men was estimated after gallbladder cancer (SMR, 3.82 [95% CI, 3.31-4.39]; mortality, 341 per 10 000 person-years), and among women, the greatest SMR was estimated after laryngeal cancer (SMR, 4.56 [95% CI, 4.11-5.06]; mortality, 268 per 10 000 person-years).

For the 3 most common FPC types, the overall SIRs in men were statistically significantly lower for prostate cancer (SIR, 0.91 [95% CI, 0.91-0.93]; incidence, 165 per 10 000 person-years), were not significantly different for colorectal cancer (SIR, 0.99 [95% CI, 0.98-1.02]; incidence, 190 per 10 000 person-years), and were statistically significantly higher for bladder cancers (SIR, 1.08 [95% CI, 1.06-1.11]; incidence, 222 per 10 000 person-years); whereas the SMR was statistically significantly lower for prostate cancer (SMR, 0.90 [95% CI, 0.89-0.91]; mortality, 94 per 10 000 person-years) and was statistically significantly higher for colorectal cancer (SMR, 1.11 [95% CI, 1.09-1.15]; mortality, 102 per 10 000 person-years) and also for bladder cancer (SMR, 1.31 [95% CI, 1.28-1.35]; mortality, 136 per 10 000 person-years). In women, the SIR and SMR were statistically significantly higher for breast cancer (SIR, 1.06 [95% CI, 1.05-1.08]; incidence, 95 per 10 000 person-years; and SMR, 1.13 [95% CI, 1.11-1.15]; mortality, 52 per 10 000 person-years), colorectal cancer (SIR, 1.04 [95% CI, 1.02-1.07]; incidence, 134 per 10 000 person-years; and SMR, 1.15 [95% CI, 1.12-1.20]; mortality, 73 per 10 000 person-years), and uterine corpus cancer (SIR, 1.03 [95% CI, 1.01-1.06]; incidence, 124 per 10 000 person-years; and SMR, 1.09, [95% CI, 1.05-1.13]; mortality, 58 per 10 000 person-years). Additionally, the risk of developing and dying from SPCs was statistically significantly lower in male survivors of melanoma (of the skin) but for none of the FPC types in women.

Figure 1 and Figure 2 show the percentage contributions of each FPC and SPC combination to the total incidence of SPCs and the SIRs with statistically significantly higher than expected values for 264 associations of the FPC with the risk of specific type of SPCs out of 1007 eligible statistical tests. The corresponding metrics for mortality are presented in Figure 3 and Figure 4 for 274 associations with higher than expected risks out of 760 eligible statistical tests. The patterns in incidence and mortality were generally similar, although the estimates for the corresponding associations were usually greater for SMRs than for SIRs (eTable 6 and eTable 7 in Supplement 1). Results for statistically significantly lower than expected SIRs and SMRs are presented in eFigure 2 in Supplement 1.

Among all survivors, the 4 SPCs that contributed most to the total incidence of SPCs were lung (19.1%), prostate (13.7%), urinary bladder (11.1%), and colorectal cancers (10.1%) in men; and lung (19.3%), breast (17.3%), colorectal (11.0%), and uterine corpus cancers (7.4%) in women (see the “All first primary cancers” row in Figure 1A and Figure 2A; eTable 8 in Supplement 1). The 4 SPCs that contributed most to the total SPC mortality were lung (33.1% in men; 31.2% in women), colorectal (8.8% in men), pancreatic (8.5% in men; 9.4% in women), followed by non-Hodgkin lymphoma (6.0%) in men, and breast (5.8%) in women (Figure 3A and Figure 4A; eTable 8 in Supplement 1).

The greatest SIR (Figure 1B) and SMR (Figure 3B) in men was observed among survivors of Kaposi sarcoma for anal cancer (SIR, 71.3 [95% CI, 56.8-88.4]; incidence, 23.2 per 10 000 person-years; and SMR, 107.5 [95% CI, 58.8-180.4]; mortality, 4.1 per 10 000 person-years). The greatest SIR in women was among survivors of laryngeal cancer (SIR, 19.3 [95% CI, 14.5-25.1]; incidence, 39.9 per 10 000 person-years) and esophageal cancer (SIR, 17.2 [95% CI, 10.2-27.2]; incidence, 39.0 per 10 000 person-years) for oral cavity/pharyngeal cancer (Figure 2B); whereas the greatest SMR in women was seen among survivors of eye and orbit cancer (SMR, 212.9 [95% CI, 176.2-254.9]; mortality, 118.4 per 10 000 person-years) for melanoma (Figure 4B).

Among survivors of 12 smoking-related cancers, the risks of developing any subsequent smoking-related cancers were statistically significantly higher for all cancers except liver cancer in men and stomach cancer and acute nonlymphocytic leukemia in women, with SIRs ranging from 1.08 (95% CI, 1.04-1.13) after female colorectal cancer to 4.62 (95% CI, 4.17-5.11) after female laryngeal cancer survivors (Figure 5A; eTable 9 in Supplement 1). Among survivors of 12 obesity-related cancers, the risks of developing any subsequent obesity-related cancers were statistically significantly higher among male survivors of pancreatic, gallbladder, thyroid, and colorectal cancers and among female survivors of thyroid, kidney, uterine corpus, and breast cancers (Figure 5B; eTable 9 in Supplement 1). Similarly, the risk for alcohol- and infection-related cancers was significantly higher after most alcohol- and infection-related cancers (eFigure 3 in Supplement 1).

Temporal trends in SIRs were statistically significant for several cancers (eFigure 4 and eTable 10 in Supplement 1). For example, the ratio of SIRs for lung cancer comparing 2000-2011 with 1975-1989 was 1.52 (95% CI, 1.43-1.61) among laryngeal cancer survivors, and 1.20 (95% CI, 1.16-1.24) among female breast cancer survivors. The corresponding ratio of the SIRs for colorectal cancer was 1.28 (95% CI, 1.20-1.36) among uterine corpus cancer survivors.

In sensitivity analyses, SIRs restricted to the first SPCs among 5-year survivors (eTable 11 in Supplement 1) or expanded to 1-year survivors (eTable 12 in Supplement 1) were generally similar to those from the main analysis regarding the directions of the associations, although the point estimates generally were lower for the former group and higher for the latter group.

Quiz Ref IDBased on the most recent population-based incidence data in the US, cancer survivors overall had higher risks of developing or dying from SPCs than the general population. The risk of developing SPCs was significantly higher for 18 of the 30 FPC types in men and for 21 of the 31 FPC types in women, with significantly higher risks of dying from SPCs for the majority of cancer types. Across FPC types, associations with specific types of SPCs varied considerably; however, only a few smoking- or obesity-related SPCs comprised substantial proportions of the total SPC incidence and mortality, highlighting opportunities for primary prevention as central components of survivorship care.

Quiz Ref IDAlthough some of the higher-risk SPCs among survivors may reflect genetic predisposition and treatment exposures, most of this excess risk is likely due to host factors (eg, aging, immunity) and to lifestyle risk factors (eg, smoking, obesity, alcohol, infection) shared by the first and subsequent cancers.²³ The risks of smoking-related SPCs were commonly elevated following many types of smoking-related FPCs, suggesting the role of smoking as a shared risk factor. Among survivors of all FPCs, 4 common smoking-related SPCs—lung, urinary bladder, oral cavity/pharynx, and esophagus—accounted for 26% to 45% of the total SPC incidence and mortality. Furthermore, lung cancer alone comprised 31% to 33% of the total mortality from SPCs. In 2018, 12% of cancer survivors in the US were current smokers,²⁷ with higher prevalence among survivors of smoking-related cancers than survivors of non–smoking-related cancers (20% vs 11%).²⁸ Only 46% of those smokers, however, reported using counseling or medication when trying to quit, and less than 8% reported using both quitting aids,²⁹ suggesting the majority of attempts to quit are inadequately supported. Thus, systematic efforts are needed to increase the provision of counseling and pharmacological interventions for tobacco dependence among survivors.

Quiz Ref IDSimilarly, survivors of many obesity-related cancers had an elevated risk of developing obesity-related SPCs. Further, among survivors of all FPCs, 4 common obesity-related cancers—colorectum, pancreas, corpus uteri, and liver—comprised 22% to 26% of total SPC mortality. Prior studies of breast and colorectal cancer survivors demonstrated greater risk of obesity-associated SPCs among those with excess body weight,^30,31 suggesting that maintenance of heathy bodyweight may reduce the risk of SPCs. In 2018, 67% of cancer survivors in the US were overweight or obese, and 34% of survivors reported no leisure time physical activity.²⁷ Many cancer survivors, however, report that they have never discussed lifestyle recommendations relevant to bodyweight with their health care providers, nor have they changed these behaviors.^32,33 Survivorship care may require greater focus on lifestyle factors, including weight management, physical activity, and healthy eating; survivorship care guidelines recommending health promotion need wider dissemination and implementation in oncology and primary care.³⁴

Infections may also play a role in the risk of SPCs. For example, co-occurrence of human papillomavirus–associated cancers (oral cavity/pharynx, anus, external genital organs, cervix) is likely related to cross-infection at multiple sites.³⁵ The large excess risk of anal cancer among male Kaposi sarcoma survivors and co-occurrence of cancers of the liver, oral cavity/pharynx, and non-Hodgkin lymphoma may reflect underlying immunosuppression, possible role of herpesvirus 8, and behavioral factors that increase chances of viral infection (eg, human immunodeficiency virus, human papillomavirus, hepatitis viruses B and C).^8,36 Primary prevention for some of these cancers is possible with vaccination against hepatitis B and treatments of hepatitis C infection and precancerous lesions (eg, advanced anal intraepithelial neoplasia).

The risks of developing SPCs among survivors compared with those in the general population increased for many FPC types between 1975-1989 and 2000-2015. Reasons for these trends are unknown but may include changes in risk factor prevalence and treatment patterns. Quiz Ref IDConsistent with previous studies, the high risks of developing and dying from acute nonlymphocytic leukemia were observed after multiple solid tumor types, likely reflecting late effects of chemotherapy, radiation, or both.²⁴ Higher risks of developing SPCs at sites in close proximity to the FPC (eg, breast cancer after Hodgkin lymphoma; bladder cancer after prostate or testicular cancer) and subsequent soft tissue sarcomas observed after several cancers likely reflect, in part, the effects of radiotherapy.³⁷ Because the use of leukemogenic agents and radiotherapy have increased over time,^37,38 quantifying the late effects of prior therapies remains highly relevant for clinicians to inform risk-based long-term follow-up care planning.

In addition to reinforcing the importance of coordinated efforts by primary care clinicians to mitigate the risks of SPCs through survivorship care— counseling to promote adherence to healthy lifestyle recommendations and surveillance and screening for new malignancies—these findings have implications for reducing the economic burden of cancer. With the growing number of long-term survivors, the costs of treating patients with multiple primary cancers will increase, amplifying financial burden for cancer survivors and their families, particularly among elderly persons who may be living on fixed incomes.³⁹ This consideration also has implications for the Medicare program, the primary payer for the population aged 65 years and older, as well as other health care payers.

This study has several limitations. First, the overall analysis may conceal important heterogeneity in the risk of SPCs by age, latency, treatment patterns, and other clinical and sociodemographic factors. Although the results were stratified by sex, formal statistical tests were not conducted for these differences. Importantly, this study did not examine risk associated with receipt of radiation and systemic treatments because this information in the registries was classified as “no/unknown” for a considerable proportion of patients (eg, no/unknown, 45% for radiotherapy and 39% for chemotherapy among breast cancer survivors). Second, the accuracy of underlying causes of death on death certificates for SPCs among cancer survivors is unknown. Third, SPCs diagnosed among survivors who migrate out of SEER registry areas are not reportable and, therefore, this risk may be underestimated. However, only 4% of survivors in the analytic samples were lost to follow-up. Fourth, the registry data used in this study cover 9% to 13% of the US population, and results may therefore not be generalizable to the entire population, although SEER is considered as the criterion standard for cancer surveillance and outcomes.⁴⁰ Fifth, SPCs in the same organ as the FPC were excluded to minimize bias due to misclassification of recurrence; therefore, the incidence rates may underestimate the true risk of SPCs that may include multicentric tumors, new primaries at the other side of paired organs, and those with different histology. Sixth, heightened medical surveillance following the initial cancer diagnosis may increase detection of SPCs, although SPCs occurring within 5 years of first cancer diagnosis were excluded to minimize this bias. Seventh, it is possible that some of the observed statistically significant associations could be due to chance because of multiple testing. Eighth, data were presented for the percentage contributions of smoking-, obesity-, alcohol-, and infections-related cancers to the burden of SPCs among survivors but not for population-attributable fractions of SPCs to these risk factors because of limited data on prevalence of risk factors and relative risks among survivors.

Among survivors of adult-onset cancers in the United States, several types of primary cancers were significantly associated with greater risk of developing and dying from an SPC, compared with the general population. Cancers associated with smoking or obesity comprised substantial proportions of overall SPC incidence and mortality among all survivors and highlight the importance of ongoing surveillance and efforts to prevent new cancers among survivors.

Corresponding Author: Ahmedin Jemal, DVM, PhD, American Cancer Society, 250 Williams St NW, Atlanta, GA 30303 (ahmedin.jemal@cancer.org).

Accepted for Publication: November 5, 2020.

Correction: This article was corrected on April 16, 2021 for incorrect data in Figure 1, Figure 2, and Figure 3.

Author Contributions: Dr Sung had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Sung, Yabroff, Jemal.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Sung, Jemal, Yabroff.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Sung, Hyun.

Administrative, technical, or material support: Jemal.

Supervision: Jemal.

Conflict of Interest Disclosures: None reported.

Funding/Support: Supported by the Intramural Research Department of the American Cancer Society (Drs Sung, Leach, Yabroff, and Jemal).

Role of the Funder/Sponsor: The American Cancer Society had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.