Analysis of temporal survival trends: considerations and best practice

Monitoring quality of healthcare is vital to ensure that changes made to clinical practice achieve the intended goals. Assessing temporal trends due to the accumulated effect of all changes in clinical practice in a given period is essential in quality monitoring. However, this assesment is compplicated by the fact that numerous of changes might occour over time unrelated to the clinical practice. Furthermore, the methods used to assess temporal trends in patient outcomes in the medical literature are heterogeneous, making it difficult to compare results between studies. In this paper, we describe methods that enable researchers to investigate temporal trends in survival data and we discuss their pros and cons. Numerous unrelated changes may occur over time which must be taken into account and disentangled when assessing the improvement in clinical management. The methods and interpretation thereof are exemplified on patients with diffuse large B-cell lymphoma from the Danish lymphoma registry.

Congestive heart failure after anthracycline-containing treatment for Hodgkin lymphoma: A Swedish matched cohort study

Introduction

Congestive heart failure (CHF) is a known complication after anthracyclines and radiotherapy for classical Hodgkin lymphoma (cHL). Contemporary cHL treatment may be associated with less risk because radiotherapy use and techniques have changed substantially over time.

Methods

In this study, Swedish cHL patients diagnosed in 2000-2018, and treated with adriamycin [doxorubicin], bleomycin, vinblastine, and dacarbazine (ABVD) or bleomycin, etoposide, Adriamycin [doxorubicin], cyclophosphamide, vincristine, procarbazine, and prednisone (BEACOPP), were matched 1:10 to the general population on birth year and sex to investigate relative rates and cumulative risks of CHF.

Results

A total of 1994 cHL patients were included, with a median age of 34 years. The median follow-up was 8.1 years. The CHF rate was higher for patients versus comparators (adjusted hazard ratio [HR] = 3.02, 95% confidence interval [CI]: 2.26-4.02). Patients treated with ≤200 mg/m2 of anthracyclines had HR of 2.89 (95% CI: 1.51-3.47) versus 3.91 (95% CI: 2.72-5.60) for >200 mg/m2. Treatment with ABVD was associated with a significantly higher CHF rate (adjusted HR = 3.25, 95% CI: 2.31-4.23), while BEACOPP was not (adjusted HR = 1.95, 95% CI: 0.91-4.16). The increase in relative rates translated to the absolute scale, with an increased risk persisting up to 18 years for low cumulative doses.

Conclusion

These findings highlight that cHL survivors still face a substantial excess risk of CHF in the modern treatment era and that focus on cardiovascular health remains relevant.

Estimating causal effects in the presence of competing events using regression standardisation with the Stata command standsurv

BACKGROUND

When interested in a time-to-event outcome, competing events that prevent the occurrence of the event of interest may be present. In the presence of competing events, various estimands have been suggested for defining the causal effect of treatment on the event of interest. Depending on the estimand, the competing events are either accommodated or eliminated, resulting in causal effects with different interpretations. The former approach captures the total effect of treatment on the event of interest while the latter approach captures the direct effect of treatment on the event of interest that is not mediated by the competing event. Separable effects have also been defined for settings where the treatment can be partitioned into two components that affect the event of interest and the competing event through different causal pathways.

METHODS

We outline various causal effects that may be of interest in the presence of competing events, including total, direct and separable effects, and describe how to obtain estimates using regression standardisation with the Stata command standsurv. Regression standardisation is applied by obtaining the average of individual estimates across all individuals in a study population after fitting a survival model.

RESULTS

With standsurv several contrasts of interest can be calculated including differences, ratios and other user-defined functions. Confidence intervals can also be obtained using the delta method. Throughout we use an example analysing a publicly available dataset on prostate cancer to allow the reader to replicate the analysis and further explore the different effects of interest.

CONCLUSIONS

Several causal effects can be defined in the presence of competing events and, under assumptions, estimates of those can be obtained using regression standardisation with the Stata command standsurv. The choice of which causal effect to define should be given careful consideration based on the research question and the audience to which the findings will be communicated.

Long-term risk of cardiovascular disease mortality among classic Hodgkin lymphoma survivors

BACKGROUND

The temporal trend of cardiovascular disease (CVD) mortality in patients with classic Hodgkin lymphoma (cHL) throughout follow-up remains unclear. This study aimed to assess this temporal trend in patients with cHL.

METHODS

This multicenter cohort included 15,889 patients with cHL diagnosed between 1983 and 2015, covering all ages. The proportional mortality ratio, cumulative incidence of cause-specific mortality accounting for competing risk, standard mortality ratio, and absolute excess risk were analyzed.

RESULTS

Among patients in stage I and stage II cHL, the proportional mortality ratio for CVD exceeded that for cHL, after approximately 60 or 120 months of follow-up, respectively. For almost all the patients with stage I or stage II disease, the cumulative incidence of CVD mortality exceeded that of cHL and other neoplasms over time. In recent decades, the risk of cHL mortality declined sharply, but the risk of CVD mortality among patients with cHL declined quite slowly or even remained unchanged among some populations. Patients with stage I or stage II disease experienced a higher risk of CVD mortality than the general population in almost all follow-up intervals. The absolute excess CVD risk among patients in stage I reached 48.5.

CONCLUSIONS

The risk of CVD mortality exceeded that of cHL and other neoplasms and became the leading cause of death over time, among patients with stage I or stage II disease. More effective measures should be taken to reduce the risk of CVD mortality.

LAY SUMMARY

Among patients with stage I and stage II classic Hodgkin lymphoma (cHL), the proportional mortality ratio of cardiovascular disease (CVD) exceeded that of cHL after approximately 60 or 120 months of follow-up, respectively. For almost all the patients with stage I or stage II disease, the cumulative incidence of CVD mortality exceeded that of cHL and other neoplasms over time. In the past several decades, the risk of cHL mortality declined sharply, but the risk of CVD mortality among patients with cHL declined quite slowly or even unchanged among some populations. CVD exceeded cHL and has become the leading cause of death over time.

Cardiologic Long-Term Follow-Up of Patients Treated With Chest Radiotherapy: When and How?

Introduction: Radiotherapy may cause valvular (VHD), pericardial, coronary artery disease (CAD), left ventricular dysfunction (LVD), arrhythmias. The risk of radiation induced heart disease (RIHD) increases over time. The current guidelines suggest a screening for RIHD every 5 years in the long-term survivors who had been treated by chest RT.

Methods: We reviewed the clinical and instrumental data of 106 patients diagnosed with RIHD. In one group (Group A: 69 patients) RIHD was diagnosed in an asymptomatic phase through a screening with ECG, echocardiogram and stress test. A second group (37 patients) was seen when RIHD was symptomatic. We compared the characteristics of the two groups at the time of RT, of RIHD detection and at last follow-up.

Results: Overall, 64 patients (60%) had CAD (associated to other RIHD in 18); 39 (36.7%) had LVD (isolated in 20); 24 (22.6%) had VHD (isolated in 10 cases). The interval between the last negative test and the diagnosis of moderate or severe RIHD was <5 years in 26 patients, and <4 years in 18. In group A, 63% of the patients with CAD had silent ischemia. The two groups did not differ with regard to type of tumor, cardiovascular risk factors, use of anthracycline-based chemotherapy, age at RT treatment, radiation dose and interval between RT and toxicity detection. The mean time from RT and RIHD was 16 years in group A and 15 in group B. Interventional therapy at RIHD diagnosis was more frequent in group B (54 vs. 30%, p < 0.05). At last follow-up, 27 patients had died (12 of cancer, 9 of cardiac causes, 6 of other causes); mean ejection fraction was 60% in group A and 50% in group B (p < 0.01). Patients with ejection fraction ≤ 50% were 14.5% in group A and 40% in group B (p < 0.01).

Conclusions: Clinically relevant RIHD become evident at a mean interval of 16 years after RT. The most frequent clinical manifestations are CAD and LVD. RIHD diagnosis in asymptomatic patients may preserve their cardiac function with timely interventions. We suggest -after 10 years from radiotherapy- a screening every 2–3 years.

Anthracycline treatment, cardiovascular risk factors and the cumulative incidence of cardiovascular disease in a cohort of newly diagnosed lymphoma patients from the modern treatment era

The development of cardiovascular disease (CVD) in long-term survivors of lymphoma is of increasing importance. Here, we characterize the cumulative incidence and risk factors for CVD in lymphoma patients diagnosed in the current treatment era. From 2002-2015, newly diagnosed lymphoma patients (>18 years) were enrollment into a prospective cohort study that captured incident CVD, consisting of congestive heart failure (CHF), acute coronary syndrome (ACS), valvular heart disease (VHD), and arrhythmia. The cumulative incidence of CVD was calculated with death modeled as a competing risk. We estimated the association of treatment with anthracyclines or radiotherapy and traditional CVD risk factors with incidence of CVD using hazard ratios (HR) and 95% confidence intervals (CI) estimated from Cox regression. After excluding prevalent CVD at lymphoma diagnosis, the study consisted of 3063 patients with a median age of 59 years (range 18-95). The cumulative incidence of CVD at 10-years was 10.7% (95% CI, 9.5%-12.1%). In multivariable analysis, increasing age (HR = 1.05 per year, p < 0.001), male sex (HR = 1.36, p = 0.02), current smoker (HR = 2.10, p < 0.001), BMI > 30 kg/m2 (HR = 1.45, p = 0.01), and any anthracycline treatment (HR = 1.57, p < 0.001) were all significantly associated with risk of CVD. Anthracyclines were associated with increased risk of CHF (HR = 2.71, p < 0.001) and arrhythmia (HR = 1.61, p < 0.01), but not VHD (HR = 0.84, p = 0.58) or ACS (HR = 1.32, p = 0.24) after adjustment for CVD risk factors. Even in the modern treatment era, CVD remains common in lymphoma survivors and preventive efforts are required that address both treatment and CVD risk factors.

A multistate model incorporating estimation of excess hazards and multiple time scales

As cancer patient survival improves, late effects from treatment are becoming the next clinical challenge. Chemotherapy and radiotherapy, for example, potentially increase the risk of both morbidity and mortality from second malignancies and cardiovascular disease. To provide clinically relevant population-level measures of late effects, it is of importance to (1) simultaneously estimate the risks of both morbidity and mortality, (2) partition these risks into the component expected in the absence of cancer and the component due to the cancer and its treatment, and (3) incorporate the multiple time scales of attained age, calendar time, and time since diagnosis. Multistate models provide a framework for simultaneously studying morbidity and mortality, but do not solve the problem of partitioning the risks. However, this partitioning can be achieved by applying a relative survival framework, allowing us to directly quantify the excess risk. This article proposes a combination of these two frameworks, providing one approach to address (1) to (3). Using recently developed methods in multistate modeling, we incorporate estimation of excess hazards into a multistate model. Both intermediate and absorbing state risks can be partitioned and different transitions are allowed to have different and/or multiple time scales. We illustrate our approach using data on Hodgkin lymphoma patients and excess risk of diseases of the circulatory system, and provide user-friendly Stata software with accompanying example code.

Cause-Specific Mortality Following Initial Chemotherapy in a Population-Based Cohort of Patients With Classical Hodgkin Lymphoma, 2000-2016

PURPOSE

Mortality for patients with classical Hodgkin lymphoma (cHL) treated during an era characterized in the United States by widespread use of doxorubicin, bleomycin, vinblastine, and dacarbazine and diminishing use of radiotherapy is not well understood.

PATIENTS AND METHODS

We identified 20,007 individuals diagnosed with stage I/II (early) or III/IV (advanced) cHL between age 20 and 74 years treated with initial chemotherapy in US population-based cancer registries during 2000-2015 (follow-up through 2016). We used standardized mortality ratios (SMRs) to compare cause-specific relative mortality risk following cHL to that expected in the general population and estimated excess absolute risks (EARs; per 10,000 patient-years) to quantify disease-specific death burden.

RESULTS

We identified 3,380 deaths in the cHL cohort, including 1,321 (39%) not attributed to lymphoma. Overall, noncancer SMRs were increased 2.4-fold (95% CI, 2.2 to 2.6; observed, 559; EAR, 61.6) and 1.6-fold (95% CI, 1.4 to 1.7; observed, 473; EAR, 18.2) for advanced- and early-stage cHL, respectively, compared with the general US population. SMRs and EARs differed substantially by cause of death and cHL stage. Among the highest EARs for noncancer causes of death were those for heart disease (EAR, 15.1; SMR, 2.1), infections (EAR, 10.6; SMR, 3.9), interstitial lung disease (ILD; EAR, 9.7; SMR, 22.1), and adverse events (AEs) related to medications/drugs (EAR, 7.4; SMR, 5.0) after advanced-stage cHL and heart disease (EAR, 6.6; SMR, 1.7), ILD (EAR, 3.7; SMR, 13.1), and infections (EAR, 3.1; SMR, 2.2) after early-stage cHL. Strikingly elevated SMRs for ILD, infections, and AEs were observed < 1 year after cHL. Individuals age 60-74 years with advanced-stage cHL experienced a disproportionate excess of deaths as a result of heart disease, ILD, infections, AEs, and solid tumors.

CONCLUSION

Despite evolving cHL treatment approaches, patients continue to face increased nonlymphoma mortality risks from multiple, potentially preventable causes. Surveillance, early interventions, and cHL treatment refinements may favorably affect patient longevity, particularly among high-risk subgroups.

Adjuvant chemotherapy administration and survival outcomes of lymphoma survivors with common solid tumors: a population-based study†

Using the Surveillance, Epidemiology, and End Results database (2004-2015), we compared adjuvant chemotherapy use and survival for three common solid tumors in patients with and without history of lymphoma (DLBCL: diffuse large B cell, HL: Hodgkin lymphoma). Among patients with breast (n = 531,243), colon (n = 108,196), and lung (n = 23,179) cancers, we identified 361, 134, and 37 DLBCL survivors, and 349, 73, and 25 HL survivors, respectively. We found no significant difference between lymphoma survivors and controls in the use of adjuvant chemotherapy, except HL survivors with colon cancer, who had a lower rate. Among chemotherapy recipients, OS was significantly worse among HL survivors with all three cancers, and DLBCL survivors with breast cancer (hazard ratio [HR] 1.57-2.28). HL survivors had significantly higher mortality from cardiovascular diseases in breast and lung cancers (sub-HR, 7.96-9.64), which suggests that worse survival in this population might be due to late or cumulative toxicities of cancer-directed treatment.

Distribution of hospital care among pediatric and young adult Hodgkin lymphoma survivors-A population-based cohort study from Sweden and Denmark

The burden of late effects among Hodgkin lymphoma (HL) survivors treated according to contemporary protocols remains poorly characterized. We used nation‐wide registers to assess number of inpatient bed‐days and specialist outpatient visits among 1048 HL‐patients (<25 years, diagnosed 1990‐2010) and 5175 country‐, sex‐, and age‐matched comparators. We followed them for up to 24 years, with time‐dependent assessment of relapse status. International Classification of Diseases (ICD‐10) chapter‐specific hazard ratios (HRs) were assessed in Cox regression analyses, and nonparametric statistics described patterns of health‐care‐use. Relative to comparators, relapse‐free survivors were at increased risk of infections, diseases of the blood, endocrine, circulatory and respiratory systems, and unspecific symptoms, HRs ranging from 1.86 to 3.05. Relative to comparators, relapsed survivors had at statistically significantly increased risk of diseases reflecting practically all investigated disease‐chapters, HRs ranging from 1.60 to 18.7. Among relapse‐free survivors, 10% of the patients accounted for 80% of all hospital bed days, and 55% were never hospitalized during follow‐up. Among relapsed‐survivors, 10% of the patients accounted for 50% of the bed days, and only 24% were never hospitalized during follow‐up. In contrast, 10% of the comparators accounted for 90% of hospital bed days and 75% were never hospitalized. These findings challenge the impression of a uniformly distributed long‐term morbidity among all HL survivors and emphasize the need for early identification and attention to patients particularly susceptible to late effects, such as relapsed survivors.

Partitioning of time trends in prevalence and mortality of lung cancer

BACKGROUND

Time trends of lung cancer prevalence and mortality are the result of three competing processes: changes in the incidence rate, stage-specific survival, and ascertainment at early stages. Improvements in these measures act concordantly to improve disease-related mortality, but push the prevalence rate in opposite directions making a qualitative interpretation difficult. The goal of this paper is to evaluate the relative contributions of these components to changes in lung cancer prevalence and mortality.

METHODS

Partitioning of prevalence and mortality trends into their components using SEER data for 1973-2013.

RESULTS

The prevalence of lung cancer increases for females and decreases for males. In 1998, the former was due to increased incidence (45%-50% of total trend), improved survival (40%-45%), and increased ascertainment at early stages (10%-15%). In males, a rapidly declining incidence rate overpowered the effects of survival and ascertainment resulting in an overall decrease in prevalence over time. Trends in lung cancer mortality are determined by incidence during 1993-2002 with noticeable contribution of survival after 2002.

CONCLUSION

Lung cancer incidence was the main driving force behind trends in prevalence and mortality. Improved survival played essential role from 2000 onwards. Trends in stage ascertainment played a small but adverse role. Our results suggest that further improvement in lung cancer mortality can be achieved through advances in early stage ascertainment, especially for males, and that in spite of success in treatment, adenocarcinoma continues to exhibit adverse trends (especially in female incidence) and its role among other histology-specific lung cancers will increase in the near future.

Relapse Risk and Loss of Lifetime After Modern Combined Modality Treatment of Young Patients With Hodgkin Lymphoma: A Nordic Lymphoma Epidemiology Group Study

PURPOSE

Estimates of short- and long-term survival for young patients with classic Hodgkin lymphoma (cHL) are of considerable interest. We investigated cHL prognosis in the era of contemporary treatment at different milestones during the follow-up.

PATIENTS AND METHODS

On the basis of a Nordic cohort of 2,582 patients diagnosed at ages 18 to 49 years between 2000 and 2013, 5-year relapse risks and 5-year restricted losses in expectation of lifetime were estimated for all patients and for patients who achieved event-free survival (EFS) for 12 (EFS12), 24 (EFS24), 36 (EFS36) or 60 (EFS60) months. The median follow-up time was 9 years (range, 2.9 to 16.8 years).

RESULTS

The 5-year overall survival was 95% (95% CI, 94% to 96%). The 5-year risk of relapse was 13.4% (95% CI, 12.1% to 14.8%) overall but decreased to 4.2% (95% CI, 3.8% to 4.6%) given that patients reached EFS24. Relapse risk for patients treated with six to eight courses of bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone (BEACOPP) was comparable to that of patients treated with six to eight courses of doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) despite more adverse risk criteria among patients treated with BEACOPP. Both from diagnosis and if EFS24 was reached, the losses in expectation of lifetime during the following 5 years were small (from diagnosis, 45 days [95% CI, 35 to 54 days] and for patients who reached EFS24, 13 days [95% CI, 7 to 20 days]). In stage-stratified analyses of 5-year restricted loss in expectation of lifetime, patients with stages I to IIA disease had no noteworthy excess risk of death after they reached EFS24, whereas risk remained measurable for patients with stages IIB to IV cHL.

CONCLUSION

Real-world data on young patients with cHL from the Nordic countries show excellent outcomes. The outlook is particularly favorable for patients who reach EFS24, which supports limited relapse-oriented clinical follow-up.

FDG-PET/CT in the management of lymphomas: current status and future directions

FDG-PET/CT is the current state-of-the-art imaging in lymphoma and plays a central role in treatment decisions. At diagnosis, accurate staging is crucial for appropriate therapy selection: FDG-PET/CT can identify areas of lymphoma missed by CT alone and avoid under-treatment of patients with advanced disease stage who would have been misclassified as having limited stage disease by CT. Particularly in Hodgkin lymphoma, positive interim FDG-PET/CT scans are adversely prognostic for clinical outcomes and can inform PET-adapted treatment strategies, but such data are less consistent in diffuse large B-cell lymphoma. The use of quantitative FDG-PET/CT metrics using metabolic tumour volume, possibly in combination with other biomarkers, may better define prognostic subgroups and thus facilitate better treatment selection. After chemotherapy, FDG-PET/CT response is predictive of outcome and may identify a subgroup who benefit from consolidative radiotherapy. Novel therapies, in particular immunotherapies, exhibit different response patterns than conventional chemotherapy, which has led to modified response criteria that take into account the risk of transient pseudo-progression. In relapsed lymphoma, FDG-PET/CT after second-line therapy and prior to high-dose therapy is also strongly associated with outcome and may be used to guide intensity of salvage therapy in relapsed Hodgkin lymphoma. Currently, FDG-PET/CT has no role in the routine follow-up after complete metabolic response to therapy, but it remains a powerful tool for excluding relapse if patients develop clinical features suggestive of disease relapse. In conclusion, FDG-PET/CT plays major roles in the various phases of management of lymphoma and constitutes a step towards the pursuit of personalized treatment.

Long-term risk of cardiovascular mortality in lymphoma survivors: A systematic review and meta-analysis

Cardiovascular disease has been identified as one of the late complications of cancer therapy. The purpose of this study was to quantify the long-term risk of cardiovascular mortality among lymphoma survivors relative to that of the general population. A systematic review and meta-analysis were conducted. Articles were identified in November 2016 by searching EMBASE, MEDLINE, and CINAHL databases. Observational studies were included if they assessed cardiovascular mortality in patients with lymphoma who survived for at least 5 years from time of diagnosis or if they had a median follow-up of 10 years. A pooled standardized mortality ratio (SMR) was estimated using a DerSimonian and Laird random-effects model. The Q and I2 statistics were used to assess heterogeneity. Funnel plots and Begg's and Egger's tests were used to evaluate publication bias. Of the 7450 articles screened, 27 studies were included in the systematic review representing 46 829 Hodgkin and 14 764 non-Hodgkin lymphoma survivors. The pooled number of deaths attributable to cardiovascular disease among Hodgkin and non-Hodgkin disease was estimated to be 7.31 (95% CI: 5.29-10.10; I2  = 95.4%) and 5.35 (95% CI: 2.55-11.24; I2  = 94.0%) times that of the general population, respectively. This association was greater among Hodgkin lymphoma survivors treated before the age of 21 (pooled SMR = 13.43; 95% CI: 9.22-19.57; I2  = 78.9%). There was a high degree of heterogeneity and a high risk of bias due to confounding in this body of literature. Lymphoma survivors have an increased risk of fatal cardiovascular events compared to the general population and should be targeted for cardiovascular screening and prevention campaigns.

Greater attention should be paid to developing therapies for elderly patients with Hodgkin lymphoma-A population-based study from Sweden

OBJECTIVE

Forty percent of Hodgkin lymphoma (HL) patients are older than 50 years at diagnosis, a fact which is not commonly recognized. Older patients do significantly worse than younger patients and are rarely included in clinical trials.

METHODS

Using data from Swedish Cancer and Lymphoma Registries, we estimated relative survival ratios (RSRs) for 7997 HL patients (diagnosed 1973-2013; 45% ≥50 years).

RESULTS

The 1-year RSRs (95% confidence interval; CI) for males aged 45-59, 60-69, 70-80, and 81 years and over, diagnosed in 2013, were 0.95 (0.91-0.97), 0.88 (0.81-0.92), 0.74 (0.63-0.81), and 0.52 (0.35-0.67), respectively. The corresponding 1-year RSRs for females were 0.97 (0.94-0.98), 0.91 (0.85-0.95), 0.82 (0.73-0.88), and 0.66 (0.50-0.77). No improvements in 1-year of 5-year relative survival from 2000 to 2013 were observed for patients aged 45-59 or 60-69 but there were modest improvements for patients aged 70 years and older. Importantly, we saw no changes in the distribution of disease or patient characteristics between 2000 and 2013.

CONCLUSIONS

Elderly patients constitute a large group with clearly unmet medical needs. Our findings motivate a more active approach to including elderly patients in clinical trials. Our study provides a baseline for outcome comparison after the broader introduction of targeted drugs.

Reduced Mortality Risk in the Recent Era in Early-Stage Hodgkin Lymphoma Patients Treated With Radiation Therapy With or Without Chemotherapy

PURPOSE

To determine the effect of treatment changes over time on all-cause mortality risk in patients with early-stage Hodgkin lymphoma (HL) after radiation therapy. The long-term survivorship of those with HL necessitates quantification of the late risk of mortality from HL and other causes.

METHODS AND MATERIALS

An institutional review board-approved retrospective study was conducted using a multi-institutional database of 1541 stage I and II HL patients treated from 1968 to 2007 with radiation therapy alone or combined-modality treatment. The analytic methods included cumulative incidence function, Kaplan-Meier estimates and log-rank tests for overall survival (OS) differences, and Cox proportional hazards modeling.

RESULTS

The median age at diagnosis was 27 years. At a median follow-up of 15.2 years (35% of patients with >20 years of follow-up), 395 patients had died of all causes, including 85 HL, 168 second malignancy (25 hematologic and 143 nonhematologic), 70 cardiovascular, and 21 pulmonary deaths. The cumulative incidence of non-HL mortality had surpassed HL mortality at 8.3 years. For patients treated from 1968 to 1982, 1983 to 1992, and 1993 to 2007, the 15-year OS rates were 78%, 85%, and 88%, respectively (P=.0016). On Cox proportional hazards analysis, age, B symptoms, and number of disease sites were significantly associated with all-cause mortality in the first decade of follow-up, with a trend toward significance for radiation field extent.

CONCLUSIONS

The all-cause mortality risk was significantly lower for patients treated in the most recent era during the first decade of follow-up, likely due to improved HL therapy resulting in a higher cure rate and lower treatment-related toxicity from smaller radiation fields. Current efforts toward radiation treatment reduction might further reduce the long-term mortality risk for these patients.

Increased healthcare use up to 10 years among relapse-free Hodgkin lymphoma survivors in the era of intensified chemotherapy and limited radiotherapy

With today's excellent cure rates for Hodgkin lymphoma (HL), the number of long-term survivors is increasing. This study aims to provide a global assessment of late adverse effects for working-age HL survivors treated with contemporary protocols (combination chemotherapy and limited radiotherapy). From Swedish nationwide registers we identified 1017 HL survivors diagnosed in 2000-2009, aged 18-60 years (median 32) and surviving at least one year post-diagnosis, and 4031 age-, sex-, and calendar-year-matched population comparators. Incidence rate ratios (IRR) and 95% confidence intervals (95%CI) for outpatient visits and inpatient bed-days after the first year up to 14 years post-diagnosis (through 2013) were estimated across treatment subgroups, considering relapse-free time and using negative binomial regression. Scheduled outpatient visits for HL were excluded. The rate of outpatient visits was nearly double (IRR = 1.8, 95%CI: 1.6-2.0) that among comparators and higher rates persisted up to 10 years post-diagnosis. The rate of inpatient bed-days among relapse-free survivors was more than three-fold (IRR = 3.6, 95%CI: 2.7-4.7) that of comparators and the increase persisted up to four years post-diagnosis. Patients requiring 6-8 chemotherapy courses had higher rates of outpatient visits (IRR = 1.4, 95%CI: 1.1-1.7) and bed-days (IRR = 4.7, 95%CI: 2.9-7.8) than patients treated with 2-4 courses + radiotherapy. Previously seldom reported reasons for the excess healthcare use included chest pain, keratitis, asthma, diabetes mellitus, and depression. Contemporary treatment, chemotherapy in particular, was associated with excess healthcare use among HL survivors during the first decade postdiagnosis. The reasons for healthcare visits reflected a wide range of disorders, indicating the need of broad individualized care in addition to specific screening programs.

A general framework for parametric survival analysis

Parametric survival models are being increasingly used as an alternative to the Cox model in biomedical research. Through direct modelling of the baseline hazard function, we can gain greater understanding of the risk profile of patients over time, obtaining absolute measures of risk. Commonly used parametric survival models, such as the Weibull, make restrictive assumptions of the baseline hazard function, such as monotonicity, which is often violated in clinical datasets. In this article, we extend the general framework of parametric survival models proposed by Crowther and Lambert (Journal of Statistical Software 53:12, 2013), to incorporate relative survival, and robust and cluster robust standard errors. We describe the general framework through three applications to clinical datasets, in particular, illustrating the use of restricted cubic splines, modelled on the log hazard scale, to provide a highly flexible survival modelling framework. Through the use of restricted cubic splines, we can derive the cumulative hazard function analytically beyond the boundary knots, resulting in a combined analytic/numerical approach, which substantially improves the estimation process compared with only using numerical integration. User-friendly Stata software is provided, which significantly extends parametric survival models available in standard software.