Top ten errors of statistical analysis in observational studies for cancer research

Poor statistical reporting: do we have a reason for concern? A narrative review and recommendations

PURPOSE OF REVIEW

The aim of the review conducted was to present recent articles indicating the need to implement statistical recommendations in the daily work of biomedical journals.

RECENT FINDINGS

The most recent literature shows an unchanged percentage of journals using specialized statistical review over 20 years. The problems of finding statistical reviewers, the impractical way in which biostatistics is taught and the nonimplementation of published statistical recommendations contribute to the fact that a small percentage of accepted manuscripts contain correctly performed analysis. The statistical recommendations published for authors and editorial board members in recent years contain important advice, but more emphasis should be placed on their practical and rigorous implementation. If this is not the case, we will additionally continue to experience low reproducibility of the research.

SUMMARY

There is a low level of statistical reporting these days. Recommendations related to the statistical review of submitted manuscripts should be followed more rigorously.

Management of liver and gastrointestinal toxicity induced by immune checkpoint inhibitors: Position statement of the AEEH-AEG-SEPD-SEOM-GETECCU

The development of the immune checkpoint inhibitors (ICI) is one of the most remarkable achievements in cancer therapy in recent years. However, their exponential use has led to an increase in immune-related adverse events (irAEs). Gastrointestinal and liver events encompass hepatitis, colitis and upper digestive tract symptoms accounting for the most common irAEs, with incidence rates varying from 2% to 40%, the latter in patients undergoing combined ICIs therapy. Based on the current scientific evidence derived from both randomized clinical trials and real-world studies, this statement document provides recommendations on the diagnosis, treatment and prognosis of the gastrointestinal and hepatic ICI-induced adverse events.

Management of liver and gastrointestinal toxicity induced by immune checkpoint inhibitors: Position statement of the AEEH-AEG-SEPD-SEOM-GETECCU

The development of the immune checkpoint inhibitors (ICI) is one of the most remarkable achievements in cancer therapy in recent years. However, their exponential use has led to an increase in immune-related adverse events (irAEs). Gastrointestinal and liver events encompass hepatitis, colitis and upper digestive tract symptoms accounting for the most common irAEs, with incidence rates varying from 2 % to 40 %, the latter in patients undergoing combined ICIs therapy. Based on the current scientific evidence derived from both randomized clinical trials and real-world studies, this statement document provides recommendations on the diagnosis, treatment and prognosis of the gastrointestinal and hepatic ICI-induced adverse events.

Is the future of peer review automated?

The rising rate of preprints and publications, combined with persistent inadequate reporting practices and problems with study design and execution, have strained the traditional peer review system. Automated screening tools could potentially enhance peer review by helping authors, journal editors, and reviewers to identify beneficial practices and common problems in preprints or submitted manuscripts. Tools can screen many papers quickly, and may be particularly helpful in assessing compliance with journal policies and with straightforward items in reporting guidelines. However, existing tools cannot understand or interpret the paper in the context of the scientific literature. Tools cannot yet determine whether the methods used are suitable to answer the research question, or whether the data support the authors' conclusions. Editors and peer reviewers are essential for assessing journal fit and the overall quality of a paper, including the experimental design, the soundness of the study's conclusions, potential impact and innovation. Automated screening tools cannot replace peer review, but may aid authors, reviewers, and editors in improving scientific papers. Strategies for responsible use of automated tools in peer review may include setting performance criteria for tools, transparently reporting tool performance and use, and training users to interpret reports.

Impact of early palliative care according to baseline symptom severity: Secondary analysis of a cluster-randomized controlled trial in patients with advanced cancer

Background

Early palliative care (EPC) improves the quality of life but may not be feasible for all patients with advanced cancer. Symptom screening has been suggested to triage patients for EPC, but scant evidence exists for this practice.

Methods

We conducted a subgroup analysis of a cluster‐randomized controlled trial of EPC vs. standard oncology care according to patients' baseline symptom scores (high [>23] vs. low [≤23] Edmonton Symptom Assessment System Distress Score [ESAS SDS]). A linear mixed‐effects model was used to account for correlation within clusters, adjusting for the baseline outcome score and all covariates in the original trial.

Results

Among the 461 participants, baseline symptom scores were high in 229 patients (127 intervention, 102 control) and low in 232 (101 intervention and 131 control). Among those with high baseline symptoms, there was improved quality of life in the EPC arm compared to controls at 4 months (adjusted difference in primary outcome of FACIT‐Sp change score [95% CI], 8.7 [2.8 to 14.5], p = 0.01; adjusted difference in QUAL‐E, 4.2 [0.9–7.5], p = 0.02); there was also improved satisfaction with care (6.9 [3.8–9.9], p = 0.001) and clinician‐patient interactions (−1.7 [−3.4 to −0.1], p = 0.04), but no significant difference in ESAS SDS (−5.6 [−12.7 to 1.4], p = 0.11). In the low baseline symptom group, there were no significant differences between arms for any outcomes.

Conclusion

EPC improved quality of life, satisfaction with care, and clinician‐patient interactions only in those with high baseline symptoms. Symptom severity may be an appropriate criterion to trigger early referrals to palliative care.

Causal Considerations Can Inform the Interpretation of Surprising Associations in Medical Registries

An exploratory analysis of registry data from 2437 patients with advanced gastric cancer revealed a surprising association between astrological birth signs and overall survival (OS) with p = 0.01. After dichotomizing or changing the reference sign, p-values <0.05 were observed for several birth signs following adjustments for multiple comparisons. Bayesian models with moderately skeptical priors still pointed to these associations. A more plausible causal model, justified by contextual knowledge, revealed that these associations arose from the astrological sign association with seasonality. This case study illustrates how causal considerations can guide analyses through what would otherwise be a hopeless maze of statistical possibilities.

How often do leading biomedical journals use statistical experts to evaluate statistical methods? The results of a survey

Scientific claims in biomedical research are typically derived from statistical analyses. However, misuse or misunderstanding of statistical procedures and results permeate the biomedical literature, affecting the validity of those claims. One approach journals have taken to address this issue is to enlist expert statistical reviewers. How many journals do this, how statistical review is incorporated, and how its value is perceived by editors is of interest. Here we report an expanded version of a survey conducted more than 20 years ago by Goodman and colleagues (1998) with the intention of characterizing contemporary statistical review policies at leading biomedical journals. We received eligible responses from 107 of 364 (28%) journals surveyed, across 57 fields, mostly from editors in chief. 34% (36/107) rarely or never use specialized statistical review, 34% (36/107) used it for 10–50% of their articles and 23% used it for all articles. These numbers have changed little since 1998 in spite of dramatically increased concern about research validity. The vast majority of editors regarded statistical review as having substantial incremental value beyond regular peer review and expressed comparatively little concern about the potential increase in reviewing time, cost, and difficulty identifying suitable statistical reviewers. Improved statistical education of researchers and different ways of employing statistical expertise are needed. Several proposals are discussed.

Anxiety, Depression, and Colorectal Cancer Survival: Results from Two Prospective Cohorts

Given the unalterable nature of most risk factors for colorectal cancer (CRC) survival (e.g., disease stage), identifying modifiable determinants is critical. We investigated whether anxiety and depression were related to CRC survival using data from the Nurses' Health Study (NHS) and Health Professional Follow-up Study (HPFS). Participants who received a CRC diagnosis and provided information about anxiety (nNHS = 335; nHPFS = 232) and depression (nNHS = 893; nHPFS = 272) within 4 years of diagnosis were included. Cox regression models estimated hazard ratios (HR) and 95% confidence intervals (CI) of overall mortality, while controlling for covariates (sociodemographics, cancer characteristics, and lifestyle factors). Pooled risk estimates were derived from fixed effects meta-analyses of the cohorts. Among 1732 CRC patients, 814 deaths occurred during the 28-year follow-up. Each 1 standard deviation increase in anxiety or depression symptoms was associated with a similar 16% higher mortality risk (anxiety: 95% CI = 1.05-1.29; depression: 95% CI = 1.07-1.26). Comparable results were observed across all sensitivity analyses (introducing a 1-year lag, restricting to CRC-related mortality, considering potential behavioral pathways) and stratified models (cancer stage, sex). Our findings suggest greater anxiety and depression symptoms can not only impede adherence to healthy habits and reduce quality of life in cancer patients but could also be a marker for accelerated CRC progression.

Image Based Data Mining Using Per-voxel Cox Regression

Image Based Data Mining (IBDM) is a novel analysis technique allowing the interrogation of large amounts of routine radiotherapy data. Using this technique, unexpected correlations have been identified between dose close to the prostate and biochemical relapse, and between dose to the base of the heart and survival in lung cancer. However, most analyses to date have considered only dose when identifying a region of interest, with confounding variables accounted for post-hoc, most often using a multivariate Cox regression. In this work, we introduce a novel method to account for confounding variables directly in the analysis, by performing a Cox regression in every voxel of the dose distribution, and apply it in the analysis of a large cohort of lung cancer patients. Our method produces three-dimensional maps of hazard for clinical variables, accounting for dose at each spatial location in the patient. Results confirm that a region of interest exists in the base of the heart where those patients with poor performance status (PS), PS > 1, have a stronger adverse reaction to incidental dose, but that the effect changes when considering other clinical variables, with patient age becoming dominant. Analyses such as this will help shape future clinical trials in which hypotheses generated by the analysis will be tested.

mHealth technology for ecological momentary assessment in physical activity research: a systematic review

OBJECTIVE

To systematically review the publications on ecological momentary assessment (EMA) relating to physical activity (PA) behavior in order to classify the methodologies, and to identify the main mHealth technology-based tools and procedures that have been applied during the first 10 years since the emergence of smartphones. As a result of this review, we want to ask if there is enough evidence to propose the use of the term "mEMA" (mobile-based EMA).

DESIGN

A systematic review according to PRISMA Statement (PROSPERO registration: CRD42018088136).

METHOD

Four databases (PsycINFO, CINALH, Medline and Web of Science Core Collection) were searched electronically from 2008 to February 2018.

RESULTS

A total of 76 studies from 297 potential articles on the use of EMA and PA were included in this review. It was found that 71% of studies specifically used "EMA" for assessing PA behaviors but the rest used other terminology that also adjusted to the inclusion criteria. Just over half (51.3%) of studies (39) used mHealth technology, mainly smartphones, for collecting EMA data. The majority (79.5%) of these studies (31 out of 39) were published during the last 4 years. On the other hand, 58.8% of studies that only used paper-and-pencil were published during the first 3 years of the 10-year period analyzed. An accelerometer was the main built-in sensor used for collecting PA behavior by means of mHealth (69%). Most of the studies were carried out on young-adult samples, with only three studies in older adults. Women were included in 60% of studies, and healthy people in 82%. The studies lasted between 1 and 7 days in 57.9%, and between three and seven assessments per day were carried out in 37%. The most popular topics evaluated together with PA were psychological state and social and environmental context.

CONCLUSIONS

We have classified the EMA methodologies used for assessing PA behaviors. A total of 71% of studies used the term "EMA" and 51.3% used mHealth technology. Accelerometers have been the main built-in sensor used for collecting PA. The change of trend in the use of tools for EMA in PA coincides with the technological advances of the last decade due to the emergence of smartphones and mHealth technology. There is enough evidence to use the term mEMA when mHealth technology is being used for monitoring real-time lifestyle behaviors in natural situations. We define mEMA as the use of mobile computing and communication technologies for the EMA of health and lifestyle behaviors. It is clear that the use of mHealth is increasing, but there is still a lot to be gained from taking advantage of all the capabilities of this technology in order to apply EMA to PA behavior. Thus, mEMA methodology can help in the monitoring of healthy lifestyles under both subjective and objective perspectives. The tendency for future research should be the automatic recognition of the PA of the user without interrupting their behavior. The ecological information could be completed with voice messages, image captures or brief text selections on the touch screen made in real time, all managed through smartphone apps. This methodology could be extended when EMA combined with mHealth are used to evaluate other lifestyle behaviors.

Statistical Accuracy in Rheumatology Research

Science and related research activities are always subject to errors. Statistical accuracy is necessary in order to overcome erroneous researching and reporting practices. The present article aimed to review the current literature on statistical errors in clinical medicine articles, and to provide rheumatologists with basic recommendations regarding the use of common statistical methods in research articles. With this purpose, PubMed/MEDLINE and Web of Science databases were searched by using relevant keywords. Data so far indicate that statistical errors are common in published articles from several disciplines of medicine. Statistics is the key element of any research activity, thus, implementing statistics at each step (hypothesis development, study design, sampling/data collection, data analysis, presentation) of every research is mandatory. In this regard, awareness of common statistical errors, basic knowledge on statistical methodology and consulting an expert in biostatistics from the beginning of the research process would be of value for rheumatologists.

Microscopically positive surgical margins and local recurrence in thyroid cancer. A meta-analysis

BACKGROUND

Microscopically positive surgical margins are a prognostic factor of recurrence in advanced thyroid carcinoma. However, information on early and completely resected thyroid tumors is scarce. Some studies do not identify any association between positive margin and local recurrence. The objective of this study was to perform a meta-analysis to measure the association of microscopically positive surgical margins and local recurrence in patients who underwent total thyroidectomy.

METHODS

Clinical trials assessing the association between microscopically positive surgical margin and local recurrence in patients with early-stage, well-differentiated thyroid carcinoma who underwent total thyroidectomy were evaluated. The outcome measured was local recurrence in the thyroid bed. A systematic review and meta-analysis was done using a random-effects model.

RESULTS

Six studies with 7696 patients were identified. Methodological quality was good, and we did not identify statistical heterogeneity or publication bias. The risk difference for microscopically positive surgical margin and local recurrence was 0% (95% CI 0 to 1).

CONCLUSION

Meta-analysis did not find a statistically significant association between microscopically positive surgical margin and local recurrence in this population. A finding of microscopically positive surgical margin in the absence of other adverse factors is not an indication for adjuvant treatment.

DNA ploidy status, S-phase fraction, and p53 are not independent prognostic factors for survival in endometrioid endometrial carcinoma FIGO stage I–III

Objectives

To assess the effects on relative survival of established and new prognostic factors in stage I–III grade 1–3 endometrioid endometrial carcinoma and in the subgroup of stage I grade 1–2.

Methods

This was a population-based, retrospective study including all women (n=1113) in the western Swedish healthcare region diagnosed with International Federation of Gynecology and Obstetrics (FIGO) stage I–III grade 1–3 endometrioid endometrial carcinoma in 2006–2011. Histology, grade, stage, and age were prospectively reported to the regional clinical and national cancer registers. DNA ploidy and S-phase fraction were analyzed by flow cytometer. S-phase fraction cut-off was set at ≥8%. Tumor biopsies were classified as diploid if there was one G0/G1 peak or the DNA index was 1.0±0.04. Overexpression of p53 as determined by immunohistochemistry was positive if strong nuclear staining was found in >30% of the neoplastic cells.

Results

Based on univariable statistical analyses we found that 5-year relative survival was significantly associated with S-phase fraction, DNA ploidy, p53, stage, grade, and age. Excess mortality for S-phase fraction ≥8%, aneuploidy, and p53 overexpression was 8, 14, and 8 and times higher, respectively. However, in a multivariable regression model, adjusted for stage, grade, and age, S-phase fraction, DNA ploidy, and p53 were not statistically independent prognostic factors (p=0.413, p=0.107, p=0.208, respectively) for 5-year relative survival in stage I–III grade 1–3 endometrioid endometrial carcinoma. In a subgroup analysis of stage I grade 1–2, aneuploidy identified a subgroup with impaired 5-year relative survival.

Conclusion

We can conclude that S-phase fraction, DNA ploidy, and p53 overexpression did not improve identification of high-risk patients by stage, grade, and age in stage I–III endometrioid endometrial carcinoma. In stage I, aneuploidy and grade 2 predicted lower relative survival rates than other variables.