Critical review and comparison of continuity correction methods: The normal approximation to the binomial distribution

Hypothesis testing and sample size considerations for the test-negative design

The test-negative design (TND) is an observational study design to evaluate vaccine effectiveness (VE) that enrolls individuals receiving diagnostic testing for a target disease as part of routine care. VE is estimated as one minus the adjusted odds ratio of testing positive versus negative comparing vaccinated and unvaccinated patients. Although the TND is related to case-control studies, it is distinct in that the ratio of test-positive cases to test-negative controls is not typically pre-specified. For both types of studies, sparse cells are common when vaccines are highly effective. We consider the implications of these features on power for the TND. We use simulation studies to explore three hypothesis-testing procedures and associated sample size calculations for case-control and TND studies. These tests, all based on a simple logistic regression model, are a standard Wald test, a continuity-corrected Wald test, and a score test. The Wald test performs poorly in both case-control and TND when VE is high because the number of vaccinated test-positive cases can be low or zero. Continuity corrections help to stabilize the variance but induce bias. We observe superior performance with the score test as the variance is pooled under the null hypothesis of no group differences. We recommend using a score-based approach to design and analyze both case-control and TND. We propose a modification to the TND score sample size to account for additional variability in the ratio of controls over cases. This work enhances our understanding of the data generating mechanism in a test-negative design (TND) and how it is distinct from that of a case-control study due to its passive recruitment of controls.

Hypothesis testing and sample size considerations for the test-negative design

The test-negative design (TND) is an observational study design to evaluate vaccine effectiveness (VE) that enrolls individuals receiving diagnostic testing for a target disease as part of routine care. VE is estimated as one minus the adjusted odds ratio of testing positive versus negative comparing vaccinated and unvaccinated patients. Although the TND is related to case-control studies, it is distinct in that the ratio of test-positive cases to test-negative controls is not typically pre-specified. For both types of studies, sparse cells are common when vaccines are highly effective. We consider the implications of these features on power for the TND. We use simulation studies to explore three hypothesis-testing procedures and associated sample size calculations for case-control and TND studies. These tests, all based on a simple logistic regression model, are a standard Wald test, a continuity-corrected Wald test, and a score test. The Wald test performs poorly in both case-control and TND when VE is high because the number of vaccinated test-positive cases can be low or zero. Continuity corrections help to stabilize the variance but induce bias. We observe superior performance with the score test as the variance is pooled under the null hypothesis of no group differences. We recommend using a score-based approach to design and analyze both case-control and TND. We propose a modification to the TND score sample size to account for additional variability in the ratio of controls over cases. This work expands our understanding of the data mechanisms of the TND.

Effect of small bowel transit time on accuracy of video capsule endoscopy in evaluating suspected small bowel bleeding

BACKGROUND

Obscure small bowel bleeding is defined as gastrointestinal bleeding (GIB) that is unidentifiable with esophagogastroduodenoscopy and a colonoscopy with video capsule endoscopy (VCE) being the next gold standard step for evaluation. Small bowel transit time (SBTT) is a metric of a VCE study that is defined as the time the capsule takes to travel through the small intestine.

AIM

To determine if SBTT within the VCE study, correlates to overall detection of obscure small bowel bleeds. Furthermore, we attempted to identify any existing correlation between SBTT and re-bleeding after a negative VCE study.

METHODS

This is a single center retrospective analysis of VCE studies performed for overt and occult GIB at Einstein Medical Center, Philadelphia, between 2015 and 2019. Inclusion criteria primarily consisted of patients 18 years or older who had a VCE study done as part of the workup for a GIB. Patients with incomplete VCEs, poor preparation, or with less than 6 mo of follow up were excluded. A re-bleeding event was defined either as overt or occult within a 6-mo timeframe. Overt re-bleeding was defined as Visible melena or hematochezia with > 2 gm/dL drop in hemoglobin defined an overt re-bleeding event; whereas an unexplained > 2 gm/dL drop in hemoglobin with no visible bleeding defined an occult re-bleed.

RESULTS

Results indicated that there was a significant and positive point biserial correlation between SBTT of 220 min and detection of a bleeding focus with a statistically significant p value of 0.008. However, the area under the curve was negligible when trying to identify a threshold time for SBTT to discriminate between risk of re-bleeding events after a negative VCE.

CONCLUSION

In terms of SBTT and association with accuracy of VCE finding a bleeding focus, 220 min was found to be adequate transit time to accurately find a bleeding focus, when present. It was found that no threshold SBTT could be identified to help predict re-bleeding after a negative VCE.

Association of Ethnicity, Sex, and Age With Cancer Diagnoses and Health Care Utilization Among Children in Inner Mongolia, China

IMPORTANCE

China is experiencing a sustained increase in childhood cancer. However, whether differences exist in disease burden by ethnicity remains unclear.

OBJECTIVE

To compare differences in cancer diagnoses and health care utilization in Inner Mongolia among children subgrouped by ethnicity (Han vs Mongolian), sex, and age.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study in Inner Mongolia, China, used data on children aged 0 to 14 years with cancer from the Inner Mongolia Regional Health Information Platform, which comprises the National Basic Medical Insurance database and the Inner Mongolia cause-of-death reporting system, from January 1, 2013, to December 31, 2019. Ethnicities analyzed included Han and Mongolian; patients of other ethnicities were not included in the analysis because of the small sample size. Cancer was broadly defined as a primary malignant tumor or hematologic cancer; benign central nervous system tumors were also included. A 2-year washout period was used to exclude prevalent cases. After diagnosis, the patients were followed up until the date of death or the end of the insured status, whichever came first.

EXPOSURES

Ethnicity (Han vs Mongolian), sex (male vs female), and age (0-4, 5-9, and 10-14 years).

MAIN OUTCOMES AND MEASURES

Crude incidence, 5-year prevalence, and survival rates at 1 year and 3 years after diagnosis; health care utilization, represented by medical costs during the first year and first 3 years after diagnosis; and hospital attendance with level (tertiary vs secondary and lower-level hospitals) and location of each unique visit.

RESULTS

From 2013 to 2019, 1 106 684 (2013), 1 330 242 (2014), 1 763 746 (2015), 2 400 343 (2016), 2 245 963 (2017), 2 901 088 (2018), and 2 996 580 (2019) children aged 0 to 14 years were registered in the NBMI database. Among the 2 996 580 children enrolled in 2019, the mean (SD) age was 6.8 (4.3) years, of whom 1 572 096 (52.5%) were male, 2 572 091 (85.8%) were Han, and 369 400 (12.3%) were Mongolian. A total of 1910 patients with cancer were identified (1048 were male [54.9%]; 1559 were Han [81.6%], and 300 were Mongolian [15.7%]). There were 764 hematologic cancers (40.0%) and 1146 solid tumors (60.0%). The overall crude incidence of cancer from 2015 to 2019 was 129.85 per million children (95% CI, 123.63-136.06), with a higher incidence among Mongolian than among Han children (155.12 [95% CI, 136.81-173.43] vs 134.39 [95% CI, 127.46-141.32]). The 5-year prevalence was 428.97 per million (95% CI, 405.52-452.42) in 2020, with a higher prevalence among Mongolian than among Han children (568.49 [95% CI, 91.62-645.36] vs 404.34 [95% CI, 379.77-428.91]). The combined 1-year (2015-2019) and 3-year (2015-2017) survival rates were 72.5% (95% CI, 67.5%-77.5%) and 66.8% (95% CI, 61.6%-71.9%), respectively. The 1-year (median [IQR], $1991 [$912-$10 181] vs $3991 [$1171-$15 425]) and 3-year (median [IQR], $2704 [$954-$13 909] vs $5375 [$1283-$22 466]) postdiagnosis costs were lower among Mongolian than among Han children. A higher proportion of Mongolian patients attended low-level hospitals (45.9% vs 17.4%).

CONCLUSIONS AND RELEVANCE

In this cohort study, Mongolian children had a higher incidence and prevalence of cancer but a lower demand for medical care, suggesting that further investigations are needed to identify mechanisms underlying ethnic disparities and ensure that care is equitable.

Growth in Proportion and Disparities of HIV PrEP Use Among Key Populations Identified in the United States National Goals: Systematic Review and Meta-analysis of Published Surveys

BACKGROUND

Pre-exposure prophylaxis (PrEP) use among populations most vulnerable to HIV as identified in the national HIV prevention goals is not fully known. This systematic review assessed trends of lifetime self-reported PrEP use and disparities among key populations.

METHODS

We used the CDC HIV/AIDS Prevention Research Synthesis cumulative database of electronic and manual searches in MEDLINE, CINAHL, EMBASE, and PsycINFO from 2000 to 2019 to identify English-language primary studies reporting PrEP use. Two reviewers independently screened citations, extracted data, and assessed the risk of bias with the modified Newcastle-Ottawa Scale. We estimated pooled proportions and crude/adjusted odds ratio.

RESULTS

We identified 95 eligible studies including 95,854 US-based survey respondents. A few studies (6.3%) focused on persons who inject drugs. In 2015-2017, men who have sex with men (MSM) had highest proportion of individuals who used PrEP over their lifetime [13.9% (95% confidence interval: 8.8 to 21.1), k (number of surveys) = 49] followed by Hispanic/Latinos [11.5 (7.1 to 18.1), 12], transgender women [11.2 (5.8 to 20.6), 5], and blacks [9.9 (8.3 to 11.8), 18]. Odds of PrEP use increased by 34%/year [odds ratio = 1.34/year (95% confidence interval: 1.09 to 1.64)] and significantly increased over time among MSM [1.53/year (1.21-1.93)] and blacks [1.44 (1.13-1.83)]. People in the Southern United States [9.9 (4.7-19.7), 8] and youth [7.3 (4.7-11.2), 8] had lower rates and did not demonstrate growth [0.94 (0.29-3.18); 0.82 (0.43-1.55)]. Odds of reporting lifetime PrEP use was twice [2.07 (1.27-3.38)] as great among MSM than non-MSM.

CONCLUSIONS

Proportions of PrEP use in published surveys have been growing, but remain low for people in the Southern United States and youth, and understudied in persons who inject drugs. Limitations include few studies in certain years, whereas strengths include a large number of respondents. Culturally tailored approaches targeting vulnerable populations are essential in increasing PrEP use to reduce disparities in HIV acquisition.