A simple method for analyzing matched designs with double controls: McNemar's test can be extended

Comparison of Emergency Room Visits and Rehospitalization for Bleeding Complications following Transurethral Procedures for the Treatment of Benign Prostatic Hyperplasia: A Population-Based Retrospective Cohort Study

Background: The postoperative bleeding complications associated with laser surgery of the prostate and transurethral resection of the prostate (TURP) were compared. Methods: We used the Taiwan National Health Insurance Research Database to conduct an observational population-based cohort study. All eligible patients who received transurethral procedures between January 2015 and September 2018 were enrolled. Patients who received laser surgery or TURP were matched at a ratio of 1:1 by using propensity score matching, and the association of these procedures with bleeding events was evaluated. Results: A total of 3302 patients who underwent elective transurethral procedures were included. The multivariable Cox regression analysis revealed that diode laser enucleation of the prostate (DiLEP) resulted in significantly higher emergency room risks within 90 days after surgery due to clot retention than the Monopolar transurethral resection of the prostate (M-TURP) (Hazard Ratio: 1.52; 95% Confidence Interval [CI], 1.06–2.16, p = 0.022). Moreover, GreenLight photovaporization of the prostate (PVP) (0.61; 95% CI, 0.38–1.00 p = 0.050) and thulium laser vaporesection of the prostate (ThuVARP) (0.67; 95% CI, 0.47–0.95, p = 0.024) resulted in significantly fewer rehospitalization due to clot retention than did M-TURP. No significant increase in blood clots were observed in patients using comedications and those with different demographic characteristics and comorbidities. Conclusions: Among the investigated six transurethral procedures for Benign prostatic hyperplasia, PVP and ThuVARP were safer than M-TURP because bleeding events and clot retention were less likely to occur, even in patients receiving anticoagulant or antiplatelet therapy. However, DiLEP and holmium laser enucleation of the prostate (HoLEP) did not result in fewer bleeding events than M-TURP.

Expression of immune checkpoint regulators, programmed death-ligand 1 (PD-L1/PD-1), cytotoxic T lymphocyte antigen 4 (CTLA-4), and indolaimine-2, 3-deoxygenase (IDO) in uterine mesenchymal tumors

BACKGROUND

Immune checkpoints including programmed death-ligand 1/programmed death-1/ (PD-L1/PD-1), cytotoxic T lymphocyte antigen 4 (CTLA-4), and indolaimine-2, 3-deoxygenase (IDO) have recently emerged as effective candidates for treatment against a range of human malignancies. We have investigated their expression in the uterine mesenchymal tumors.

METHODS

Sixty-eight mesenchymal tumors were categorized into 6 diagnostic groups. We assessed PD-L1, PD-1, CTLA-4, and IDO expression on paraffin embedded tissue blocks of the uterine tumors using the respective antibodies. Immunohistochemical (IHC) stains were classified as positive when the reactions were present in at least 1% of the cell membranes for PD-L1/PD-1 or in cytoplasm for CTLA-4 and IDO, regardless of intensity. Student's t-test and McNemar's chi-square tests were carried out to analyze the results.

RESULTS

The mesenchymal neoplasms had expressed the immune checkpoints in the tumor and/or the lymphoid cells at the rate of 49% and 54% respectively. The tumor cells were positive in 10 (18%, PD-L1), 0 (0%, PD-1), 18 (32%, CTLA-4), and 13 (23%, IDO) cases while the infiltrating lymphoid cells were positive in 10 (18%, PD-L1), 23 (40%, PD-1), 18 (32%, CTLA-4), and 13 (23%, IDO) cases. Overall, comparison of paired tumor vs lymphoid cells resulted in p-values of ≤ 0.04.

CONCLUSIONS

Nearly 50% of the uterine tumors express at least one of the immune checkpoints in tumor and/or the infiltrating lymphoid cells. However, expression of the proteins in the two cellular components are mutually exclusive. Namely, when tumor cells express an immune checkpoint, the infiltrating lymphoid cells do not, and vice versa. Since the leiomyosarcomas are reportedly resistant to the immunotherapy when PD-L1 is expressed in the tumor cells, it can be posited that presence of the IHC positive lymphoid cells may be a better indicator of response to the treatment.

Risks of Serious Injury with Testosterone Treatment

BACKGROUND

Injury causes significant morbidity and mortality that is sometimes attributed to testosterone and violence. We hypothesized that prescribed testosterone might be associated with the subsequent risk of serious injury.

METHODS

We conducted a self-matched individual-patient exposure-crossover analysis comparing injury risks before and after initiation of testosterone. We selected adults treated with testosterone in Ontario, Canada, from October 1, 2012, to October 1, 2017 (enrollment) and continued until October 1, 2018 (follow-up). The primary outcome was defined as an acute traumatic event that required emergency medical care.

RESULTS

A total of 64,386 patients were treated with testosterone of whom 89% were men with a median age of 52 years. We identified 34,439 serious injuries during the baseline interval before starting testosterone (584 per month) and 7349 serious injuries during the subsequent interval after starting testosterone (565 per month). Rates of injuries were substantially above the population norm in both intervals with no significant increased risk after starting testosterone (relative risk = 1.00; 95% confidence interval: 0.96-1.04, P = 0.850). The unchanged risk extended to diverse patients, was observed for different formulations and applied to all injury mechanisms. In contrast, testosterone treatment was associated with a 48% increased risk of a thromboembolic event (relative risk = 1.48; 95% confidence interval: 1.25-1.74, P < 0.001).

CONCLUSIONS

Testosterone treatment was associated with a substantial baseline risk of serious injury that did not increase further after starting therapy. Physicians prescribing testosterone could consider basic safety reminders to mitigate injury risks.

An approach to explore for a sweet spot in randomized trials

OBJECTIVE

The objective of the study was to demonstrate how a conventional randomized trial can be analyzed through a stratified or a matched approach to identify a potential sweet spot where observed differences might be accentuated in the mid range of disease severity.

DESIGN AND SETTING

We review a landmark randomized trial of heart failure patients that tested whether implantable defibrillators reduce mortality (n = 2,521).

RESULTS

Overall, 22% (182/829) of the patients in the defibrillator group died compared with 29% (484/1,692) of patients in the control group. Proportional hazards analysis yielded a modest 25% survival benefit (hazard ratio = 0.75, 95% confidence interval: 0.63 to 0.89). Stratified analysis of the trial yielded a larger 52% survival benefit for those in the middle quintile of disease severity (hazard ratio = 0.48, 95% confidence interval: 0.29 to 0.79). In contrast, little of the survival benefit was explained by patients with the greatest disease severity (hazard ratio = 0.89, 95% confidence interval: 0.69 to 1.15). The discrepancy between crude and stratified analyses could be visualized by graphical displays and replicated with matched comparisons.

CONCLUSION

Our approach for analyzing a randomized trial could help identify a potential sweet spot of an accentuated treatment effect.

Full Moon and Out-of-Hospital Cardiac Arrest in Japan - Population-Based, Double-Controlled Case Series Analysis

Background: It is commonly believed that a full moon affects human behavior or the occurrence and outcome of various diseases; thus, the occurrence of out-of-hospital cardiac arrest (OHCA) might increase during full moon nights.

Methods and Results: This nationwide, population-based observational study consecutively enrolled OHCA patients in Japan with attempted resuscitation between 2005 and 2016. The primary outcome measure was the occurrence of OHCA. Based on the double-control method, assuming Poisson sampling, we evaluated the average number of OHCA events that occurred on full moon nights compared with that which occurred on control nights, which included events that occurred on the same calendar days 1 week before and after the full moon nights. A total of 29,552 OHCA that occurred on 148 full moon nights and 58,707 OHCA that occurred on 296 control nights were eligible for analysis. The occurrence of OHCA did not differ between full moon and control nights (199.7 vs. 198.3 per night; relative risk [RR], 1.007; 95% CI: 0.993–1.021). On subgroup analysis, compared with control nights, the RR of OHCA occurrence were 1.013 (95% CI: 0.994–1.032, P=0.166) and 0.998 (95% CI: 0.977–1.020, P=0.866) for cardiac and non-cardiac origins, respectively.

Conclusions: In this population, there was no significant difference in OHCA occurrence between full moon and control nights.

Life-threatening alcohol-related traffic crashes in adverse weather: a double-matched case-control analysis from Canada

IMPORTANCE

Drunk driving is a major cause of death in North America, yet physicians rarely counsel patients on the risks of drinking and driving.

OBJECTIVE

To test whether the risks of a life-threatening alcohol-related traffic crash were further accentuated by adverse weather.

DESIGN

Double matched case-control analysis of hospitalised patients.

SETTING

Canada's largest trauma centre between 1 January 1995 and 1 January 2015.

PARTICIPANTS

Patients hospitalised due to a life-threatening alcohol-related traffic crash.

EXPOSURE

Relative risk of a crash associated with adverse weather estimated by evaluating the weather at the place and time of the crash (cases) compared with the weather at the same place and time a week earlier and a week later (controls).

RESULTS

A total of 2088 patients were included, of whom the majority were drivers injured at night. Adverse weather prevailed among 312 alcohol-related crashes and was significantly more frequent compared with control circumstances. The relative risk of a life-threatening alcohol-related traffic crash was 19% higher during adverse weather compared with normal weather (95% CI: 5 to 35, p=0.006). The absolute increase in risk amounted to 43 additional crashes, extended to diverse groups of patients, applied during night-time and daytime, contributed to about 793 additional patient-days in hospital and was distinct from the risks for drivers who were negative for alcohol.

CONCLUSIONS

Adverse weather was associated with an increased risk of a life-threatening alcohol-related traffic crash. An awareness of this risk might inform warnings to patients about traffic safety and counselling alternatives to drinking and driving.

Individually-matched etiologic studies: classical estimators made new again

With greater access to regression-based methods for confounder control, the etiologic study with individual matching, analyzed by classical (calculator) methods, lost favor in recent decades. This design was costly, and the data sometimes mis-analyzed. Now, with Big Data, individual matching becomes an economical option. To many, however, conditional logistic regression, commonly used to estimate the incidence density ratio parameter, is somewhat of a black box whose output is not easily checked. An epidemiologist-statistician pair recently proposed a new estimator that is easily applied to data from individually-matched series with a 2:1 ratio (and no other confounding variables) using just a hand calculator or spreadsheet. Surprisingly-or possibly not-they overlooked classical estimators developed in earlier decades. This prompts me to re-introduce some of these, to highlight their considerable flexibility and ease of use, and to update them. Nowadays, for any matching ratio (M:1), the Maximum Likelihood result can be easily computed from data gathered under the matched design in two different ways, each using just the summary data. One is via any binomial regression program that allows offsets, applied to just M 'rows' of data. The other is by hand! The aim of this note is not to save on computation; instead, it is to make connections between classical and regression-based methods, to promote terminology that reflects the concepts and structure of the etiologic study, and to focus attention on what parameter is being estimated.

Methods for analyzing matched designs with double controls: excess risk is easily estimated and misinterpreted when evaluating traffic deaths

OBJECTIVES

To demonstrate analytic approaches for matched studies where two controls are linked to each case and events are accumulating counts rather than binary outcomes. A secondary intent is to clarify the distinction between total risk and excess risk (unmatched vs. matched perspectives).

STUDY DESIGN AND SETTING

We review past research testing whether elections can lead to increased traffic risks. The results are reinterpreted by analyzing both the total count of individuals in fatal crashes and the excess count of individuals in fatal crashes, each time accounting for the matched double controls.

RESULTS

Overall, 1,546 individuals were in fatal crashes on the 10 election days (average = 155/d), and 2,593 individuals were in fatal crashes on the 20 control days (average = 130/d). Poisson regression of total counts yielded a relative risk of 1.19 (95% confidence interval: 1.12-1.27). Poisson regression of excess counts yielded a relative risk of 3.22 (95% confidence interval: 2.72-3.80). The discrepancy between analyses of total counts and excess counts replicated with alternative statistical models and was visualized in graphical displays.

CONCLUSION

Available approaches provide methods for analyzing count data in matched designs with double controls and help clarify the distinction between increases in total risk and increases in excess risk.

The full moon and motorcycle related mortality: population based double control study

OBJECTIVE

To test whether a full moon contributes to motorcycle related deaths.

DESIGN

Population based, individual level, double control, cross sectional analysis.

SETTING

Nighttime (4 pm to 8 am), United States.

PARTICIPANTS

13 029 motorcycle fatalities throughout the United States, 1975 to 2014 (40 years).

MAIN OUTCOME MEASURE

Motorcycle fatalities during a full moon.

RESULTS

13 029 motorcyclists were in fatal crashes during 1482 relevant nights. The typical motorcyclist was a middle aged man (mean age 32 years) riding a street motorcycle with a large engine in a rural location who experienced a head-on frontal impact and was not wearing a helmet. 4494 fatal crashes occurred on the 494 nights with a full moon (9.10/night) and 8535 on the 988 control nights without a full moon (8.64/night). Comparisons yielded a relative risk of 1.05 associated with the full moon (95% confidence interval 1.02 to 1.09, P=0.005), a conditional odds ratio of 1.26 (95% confidence interval 1.17 to 1.37, P<0.001), and an absolute increase of 226 additional deaths over the study interval. The increase extended to diverse types of motorcyclists, vehicles, and crashes; was accentuated during a supermoon; and replicated in analyses from the United Kingdom, Canada, and Australia.

CONCLUSION

The full moon is associated with an increased risk of fatal motorcycle crashes, although potential confounders cannot be excluded. An awareness of the risk might encourage motorcyclists to ride with extra care during a full moon and, more generally, to appreciate the power of seemingly minor distractions at all times.

Life-threatening motor vehicle crashes in bright sunlight

Bright sunlight may create visual illusions that lead to driver error, including fallible distance judgment from aerial perspective. We tested whether the risk of a life-threatening motor vehicle crash was increased when driving in bright sunlight.This longitudinal, case-only, paired-comparison analysis evaluated patients hospitalized because of a motor vehicle crash between January 1, 1995 and December 31, 2014. The relative risk of a crash associated with bright sunlight was estimated by evaluating the prevailing weather at the time and place of the crash compared with the weather at the same hour and location on control days a week earlier and a week later.The majority of patients (n = 6962) were injured during daylight hours and bright sunlight was the most common weather condition at the time and place of the crash. The risk of a life-threatening crash was 16% higher during bright sunlight than normal weather (95% confidence interval: 9-24, P < 0.001). The increased risk was accentuated in the early afternoon, disappeared at night, extended to patients with different characteristics, involved crashes with diverse features, not apparent with cloudy weather, and contributed to about 5000 additional patient-days in hospital. The increased risk extended to patients with high crash severity as indicated by ambulance involvement, surgical procedures, length of hospital stay, intensive care unit admission, and patient mortality. The increased risk was not easily attributed to differences in alcohol consumption, driving distances, or anomalies of adverse weather.Bright sunlight is associated with an increased risk of a life-threatening motor vehicle crash. An awareness of this risk might inform driver education, trauma staffing, and safety warnings to prevent a life-threatening motor vehicle crash.

LEVEL OF EVIDENCE

Epidemiologic Study, level III.