Collider and reporting biases involved in the analyses of cause of death associations in death certificates: an illustration with cancer and suicide

BACKGROUND

Mortality data obtained from death certificates have been studied to explore causal associations between diseases. However, these analyses are subject to collider and reporting biases (selection and information biases, respectively). We aimed to assess to what extent associations of causes of death estimated from individual mortality data can be extrapolated as associations of disease states in the general population.

METHODS

We used a multistate model to generate populations of individuals and simulate their health states up to death from national health statistics and artificially replicate collider bias. Associations between health states can then be estimated from such simulated deaths by logistic regression and the magnitude of collider bias assessed. Reporting bias can be approximated by comparing the estimates obtained from the observed death certificates (subject to collider and reporting biases) with those obtained from the simulated deaths (subject to collider bias only). As an illustrative example, we estimated the association between cancer and suicide in French death certificates and found that cancer was negatively associated with suicide. Collider bias, due to conditioning inclusion in the study population on death, increasingly downwarded the associations with cancer site lethality. Reporting bias was much stronger than collider bias and depended on the cancer site, but not prognosis.

RESULTS

The magnitude of the biases ranged from 1.7 to 9.3 for collider bias, and from 4.7 to 64 for reporting bias.

CONCLUSIONS

These results argue for an assessment of the magnitude of both collider and reporting biases before performing analyses of cause of death associations exclusively from mortality data. If these biases cannot be corrected, results from these analyses should not be extrapolated to the general population.

Suicidal risk associated with finasteride versus dutasteride among men treated for benign prostatic hyperplasia: nationwide cohort study

Finasteride, a 5α-reductase inhibitor used in benign prostatic hyperplasia and androgenetic alopecia, has been associated with an increased suicidal risk, whereas it is unclear whether such risk is similar to that for another 5α-reductase inhibitor, dutasteride. We aimed to assess the risk of suicidal behaviours with finasteride relative to dutasteride. A nationwide cohort study was conducted using the French National Health Data System (SNDS). Men aged 50 years or older initiating finasteride 5 mg or dutasteride 0.5 mg in France between 01-01-2012 and 30-06-2016 were included and followed until outcome (suicide death identified from death certificate or self-harm hospitalisation), treatment discontinuation or switch, death, or 31-12-2016. Self-harm by violent means or resulting in admission to an intensive care unit were also examined. Cox proportional hazards models controlled for age and psychiatric and non-psychiatric conditions by inverse probability of treatment weighting (IPTW). Analyses were stratified according to psychiatric history. The study compared 69,786 finasteride new users to 217,577 dutasteride new users (median age: 72.0 years [Q1-Q3 = 64.5-80.2] vs. 71.1 [Q1-Q3 = 65.0-79.2]). During follow-up, 18 suicide deaths (0.57/1000 person-years) and 34 self-harm hospitalisations (1.08/1000) occurred among finasteride users versus 47 deaths (0.43/1000) and 87 hospitalisations (0.79/1000) among dutasteride users. Overall, finasteride was not associated with an increased risk of any suicidal outcome (IPTW-adjusted Hazard Ratio = 1.21 [95% Confidence Interval  .87-1.67]), suicide death or self-harm hospitalisation. However, among individuals with a history of mood disorders, finasteride was associated with an increased risk of any suicidal outcome (25 versus 46 events; HR = 1.64 [95% CI 1.00-2.68]), suicide death (8 versus 10 events; HR = 2.71 [95% CI 1.07-6.91]), self-harm by violent means (6 versus 6 events; HR = 3.11 [95% CI 1.01-9.61]), and self-harm with admission to an intensive care unit (7 versus 5 events; HR = 3.97 [95% CI 1.26-12.5]). None of these risks was significantly increased among individuals without a psychiatric history. These findings do not support an increased risk of suicide with finasteride used in the treatment of benign prostatic hyperplasia. However, an increased risk cannot be excluded among men with a history of mood disorder, but this result based on a limited number of events should be interpreted with caution.

Insulin allergy: a diagnostic and therapeutic strategy based on a retrospective cohort and a case-control study

AIMS/HYPOTHESIS

Insulin allergy is a rare but significant clinical challenge. We aimed to develop a management workflow by (1) validating clinical criteria to guide diagnosis, based on a retrospective cohort, and (2) assessing the diagnostic performance of confirmatory tests, based on a case-control study.

METHODS

In the retrospective cohort, patients with suspected insulin allergy were classified into three likelihood categories according to the presence of all (likely insulin allergy; 26/52, 50%), some (possible insulin allergy; 9/52, 17%) or none (unlikely insulin allergy; 17/52, 33%) of four clinical criteria: (1) recurrent local or systemic immediate or delayed hypersensitivity reactions; (2) reactions elicited by each injection; (3) reactions centred on the injection sites; and (4) reactions observed by the investigator (i.e. in response to an insulin challenge test). All underwent intradermal reaction (IDR) tests. A subsequent case-control study assessed the diagnostic performance of IDR, skin prick and serum anti-insulin IgE tests in ten clinically diagnosed insulin allergy patients, 24 insulin-treated non-allergic patients and 21 insulin-naive patients.

RESULTS

In the retrospective cohort, an IDR test validated the clinical diagnosis in 24/26 (92%), 3/9 (33%) and 0/14 (0%) likely, possible and unlikely insulin allergy patients, respectively. In the case-control study, an IDR test was 80% sensitive and 100% specific and identified the index insulin(s). The skin prick and IgE tests had a marginal diagnostic value. Patients with IDR-confirmed insulin allergy were treated using a stepwise strategy.

CONCLUSIONS/INTERPRETATION

Subject to validation, clinical likelihood criteria can effectively guide diabetologists towards an insulin allergy diagnosis before undertaking allergology tests. An IDR test shows the best diagnostic performance. A progressive management strategy can subsequently be implemented. Continuous subcutaneous insulin infusion is ultimately required in most patients.

CLINICALTRIALS

gov: NCT01407640.

Risk of intracranial meningioma with three potent progestogens: a population-based case-control study

Background and purpose

A dose‐dependent association between the use of cyproterone acetate (CPA) and intracranial meningioma has been identified but data for other potent progestogens are scarce. The association was assessed between intracranial meningioma surgery and exposure to three potent progestogens: CPA (≥25 mg/day), nomegestrol acetate (NOMAC) (3.75–5 mg/day) and chlormadinone acetate (CMA) (2–10 mg/day).

Methods

In this nationwide population‐based case–control study, cases underwent surgery for intracranial meningioma in France from 2009 to 2018. They were matched to five control subjects for sex, year of birth and area of residence. Progestogen exposure was defined as progestogen use within the year before surgery for cases or the same date for their controls.

Results

In total, 25,216 cases were included (75% women, median age 58 years). Progestogen exposure was noted for 9.9% of cases (2497/25,216) and 1.9% (2382/126,080) of controls, with an odds ratio (OR) of 6.7 (95% confidence interval [CI] 6.3–7.1). The OR was 1.2 (1.0–1.4) for short‐term use (<1 year) and 9.5 (8.8–10.2) for prolonged use. A strong association was identified for prolonged use of CPA (OR = 22.7, 95% CI 19.5–26.4), NOMAC (OR = 6.5, 95% CI 5.8–7.2) and CMA (OR = 4.7, 95% CI 4.5–5.3). Progestogen exposure increased the risk of meningioma for all histological grades and anatomical sites, particularly for the anterior and middle skull base: OR = 35.7 (95% CI 26.5–48.2) and 23.9 (95% CI 17.8–32.2) for CPA. The estimated number of attributable cases was 2124 (95% CI 2028–2220) (212/year).

Conclusion

A strong association between prolonged exposure to potent progestogens and surgery for meningioma was observed. The risk increased from CMA to NOMAC to CPA. Individuals should be informed of this risk.

This study highlights a strong association between prolonged use of nomegestrol and chlormadinone acetate (two potent progestogens) and intracranial meningioma, although weaker than that of cyproterone acetate. The estimated number of cases was higher than 2000 in France over 10 years.

Factors associated with tuberculosis-associated haemophagocytic syndrome: a multicentre case-control study

SETTING: Tuberculosis (TB) is a potential trigger of haemophagocytic syndrome (HS) but little is known about the features of TB-associated HS.OBJECTIVE: To assess the risk factors associated with HS in patients with TB.DESIGN: We performed a multicentre case-control study assessing the medical records of adult patients diagnosed with proven TB with (TB/HS+) or without (TB/HS-) associated HS.RESULTS: Twenty-one patients with TB/HS+ (24% women, median age, 37 years [IQR 30-48]) were included in the study. Eleven patients (52%) were infected with human immunodeficiency virus and seven patients (33%) were immunocompromised due to other reasons. TB was disseminated in 17 patients (81%). Compared with 50 control TB patients (TB/HS-), patients with TB/HS+ were more likely to be immunocompromised (86% vs. 18%; P < 0.001) and to present with disseminated TB (80% vs. 12%; P < 0.001). The outcome was poorer in patients with TB/HS+, with a higher admission rate to intensive care (71% vs. 0%; P < 0.001) and a higher risk of death (38% vs. 7%; P = 0.005).CONCLUSION: TB/HS+ occurred more likely in immunocompromised patients and severely impaired the prognosis of TB. Further studies are needed to devise therapeutic strategies for patients with TB/HS+.

Contacts with health services during the year prior to suicide death in France (2013-2015)

Background

This study was designed to describe contacts with health services during the year before suicide death in France, and to compare the prevalent mental and physical conditions in these people to those of the general population.

Methods

Data were extracted from the French National Health Data System (SNDS), which comprises comprehensive claims data for inpatient and outpatient care linked to the national causes-of-death registry. Individuals, national health insurance general scheme beneficiaries (i.e. 76% of the population living in France), aged 15 years or older, who died from suicide in France in 2013-2015 were included. Medical consultations, emergency room visits, and hospitalisations during the year preceding death were collected. Conditions were identified, and standardised prevalence ratios (SPRs) were estimated to compare prevalence rates in suicide decedents with those of the general population.

Results

The study included 19,144 suicide decedents. Overall, 8.5% of suicide decedents consulted a physician or attended an emergency room on the day of death, 34.1% during the week before death, 60.9% during the month before death. Most contacts involved a general practitioner or an emergency room (46.2% of suicide decedents consulted a general practitioner during the month before death, 16.7% attended an emergency room). During the month preceding suicide, 24.4% of individuals were hospitalised at least once. Mental conditions (36.8% of cases) were 7.9-fold (SPR 95% CI: 7.7-8.1) more prevalent in suicide decedents than in the general population. The highest SPRs among physical conditions were for liver/pancreatic diseases (SPR=3.3, 95% CI: 3.1-3.6) and epilepsy (SPR=2.7, 95% CI: 2.4-3.0).

Conclusions

General practitioners and emergency departments have frequent contacts with suicide decedents during the last weeks before death and are at the forefront of suicide risk identification and prevention in individuals with mental, but also physical conditions.

Key messages

Mental and physical conditions are more common among suicide decedents than in the general population, and contacts with primary care services are frequent in the last weeks prior to suicide. Primary care services (general practitioners and emergency rooms) should be targeted for suicide preventive interventions.

Contacts with Health Services During the Year Prior to Suicide Death and Prevalent Conditions A Nationwide Study

BACKGROUND

This study was designed to describe contacts with health services during the year before suicide death in France, and prevalent mental and physical conditions.

METHODS

Data were extracted from the French National Health Data System (SNDS), which comprises comprehensive claims data for inpatient and outpatient care linked to the national causes-of-death registry. Individuals aged ≥15 years who died from suicide in France in 2013-2015 were included. Medical consultations, emergency room visits, and hospitalisations during the year preceding death were collected. Conditions were identified, and standardised prevalence ratios (SPRs) were estimated to compare prevalence rates in suicide decedents with those of the general population.

RESULTS

The study included 19,144 individuals. Overall, 8.5% of suicide decedents consulted a physician or attended an emergency room on the day of death, 34.1% during the week before death, 60.9% during the month before death. Most contacts involved a general practitioner or an emergency room. During the month preceding suicide, 24.4% of individuals were hospitalised at least once. Mental conditions (36.8% of cases) were 7.9-fold more prevalent in suicide decedents than in the general population. The highest SPRs among physical conditions were for liver/pancreatic diseases (SPR=3.3) and epilepsy (SPR=2.7).

LIMITATIONS

The study population was restricted to national health insurance general scheme beneficiaries (76% of the population living in France).

CONCLUSIONS

Suicide decedents have frequent contacts with general practitioners and emergency departments during the last weeks before death. Improving suicide risk identification and prevention in these somatic healthcare settings is needed.

Factors associated with mechanical and systemic adverse events after colonoscopy (France, 2010-2015)

Background

More than one million colonoscopies are performed every year in France. They are associated with risks of mechanical and systemic serious adverse events (SAEs) which can be associated with patient, procedure, endoscopist, and facility characteristics. We tried to identify the factors associated with colonic perforation, gastrointestinal bleeding, splenic injury, shock, myocardial infarction, stroke, pulmonary embolism, acute renal failure, and urolithiasis after colonoscopy.

Methods

We analysed data from the French national claims databases (SNDS). A total of 4,088,799 patients, 30 years or older, undergoing a first screening or diagnostic colonoscopy between 2010 and 2015 were identified. SAE rates were estimated, and risk factors associated with SAEs were identified using multilevel logistic regression models, adjusted for patient, colonoscopy, endoscopist, and facility characteristics.

Results

Increasing age was associated with an increasing incidence of mechanical and systemic SAEs. Cancer and cardiovascular comorbidities were associated with mechanical SAEs, and a higher number of pre-existing conditions was associated with shock and acute renal failure. Polypectomy, especially of polyps larger than 1 cm, was associated with an increased risk of perforation (OR = 4.1; 95% CI, 3.4-5.0) and bleeding (OR = 13.3; 95% CI, 11.7-15.1). Mechanical SAEs were associated with the endoscopist’s experience, while systemic SAEs were more frequent in public hospitals than in private clinics.

Conclusions

SAEs related to colonoscopy were more frequent in older patients and in those with comorbidities. Mechanical SAEs were more frequent when colonoscopy was performed by less experienced endoscopists. Systemic SAEs were more frequent in public hospitals, reflecting patient selection processes. The risk of both mechanical and systemic SAEs should be taken into account when deciding to perform colonoscopy, particularly in older patients with multiple pre-existing conditions.

Key messages

Systemic SAEs are not uncommon after colonoscopy and, together with intestinal SAEs, should be considered when considering the need for colonoscopy. Patients at risk of SAEs should be identified and colonoscopy should be performed by experienced endoscopists in these patients. Less invasive alternatives should also be considered in these patients.

Patient, Procedure, and Endoscopist Risk Factors for Perforation, Bleeding, and Splenic Injury After Colonoscopies

BACKGROUND & AIMS

We investigated perforations, bleeding, and splenic injuries after screening or diagnostic colonoscopies to identify patient-, procedure-, endoscopist-, and facility-associated risk factors.

METHODS

We analyzed data from the SNIIRAM-PMSI national claims databases in France. A total of 4,088,799 patients, 30 years or older, undergoing a first screening or diagnostic colonoscopy from 2010 through 2015 were identified. Rates of severe adverse events (SAEs) were estimated using stringent and broad definitions. Risk factors associated with perforations and major bleeding were estimated using multilevel logistic regression models, adjusted for patient, colonoscopy, and endoscopist characteristics.

RESULTS

Perforation rates ranged from 3.5 (stringent definition) to 7.3 (broad definition) per 10,000 procedures, bleeding rates ranged from 6.5 to 23.1 per 10,000 procedures, and splenic injury rates ranged from 0.20 to 0.34 per 10,000 procedures. Rates of 30-day mortality were 13.2 per 1000 bleeds, 29.2 per 1000 perforations, and 36.1 per 1000 splenic injuries (stringent definitions). Patient characteristics associated with SAEs were increasing age (especially for perforation), cancer, and cardiovascular comorbidities. Procedure characteristics associated with SAEs included polypectomy-especially of polyps larger than 1 cm with an increased risk of perforation (odds ratio, 4.1; 95% CI, 3.4-5.0) and bleeding (odds ratio, 13.3; 95% CI, 11.7-15.1). Less-experienced endoscopists and endoscopists who performed a smaller number of colonoscopies were independently associated with a risk of SAEs.

CONCLUSION

In an analysis of national claims databases in France, we found SAEs related to screening and diagnostic colonoscopies to be more frequent in older patients, in patients with comorbidities, and with less-experienced endoscopists. Patients at risk of SAE should be identified and colonoscopies should be performed or supervised by experienced endoscopists.

Diversity and combinations of infectious agents in 38 adults with an infection-triggered reactive haemophagocytic syndrome: a multicenter study

Reactive haemophagocytic syndrome (HS) is a rare condition that occurs in patients with infections, haematological malignancies or autoimmune diseases. Although various microorganisms are thought to trigger HS, most of the literature data on this topic have been gathered in single-centre case series. Here, we sought to characterize infectious triggers in a large, multicentre cohort of patients with HS. Patients were included in the present study if HS was solely due to one or more infections. Detailed microbiological data were recorded. Of the 162 patients with HS in the cohort, 40 (25%) had at least one infection and 38 of the latter (including 14 women, 36.8%) were included. The median age was 46 years. Seven patients were presumed to be immunocompetent (18.4%), whereas 19 patients (50%) were infected with human immunodeficiency virus and 12 patients (31.6%) were immunocompromised for other reasons. Twenty-seven patients (71.1%) had a single infection, whereas six (15.8%) and five (13.1%) patients had, respectively, two and three concomitant infections. We observed pyogenic bacterial infections (n = 7), tuberculosis (n = 10), non-tuberculous mycobacteriosis (n = 3), viral infections (n = 17: 11 cytomegalovirus, three Epstein-Barr virus, two human herpesvirus 8, one herpes simplex virus 2), parasitic infections (n = 8: four disseminated toxoplasmosis, one leishmaniasis, three malaria), fungal infections (n = 5: four pulmonary pneumocystosis and one candidaemia). Eighteen patients (47.4%) received corticosteroids and/or etoposide. Twelve patients died (31.6%). All multiple infections and all deaths occurred in immunocompromised patients. When compared with patients suffering from malignancy-associated HS, patients with infection-triggered HS were younger and more likely to be immunocompromised, and had a better outcome.

Factors associated with failure to return for HIV test results in a free and anonymous screening centre

Free and anonymous screening centres (CDAG: Centres de Depistage Anononyme et Gratuit) are public facilities set up for HIV infection diagnosis in France. Some people visiting CDAG fail to return for test results and are not informed of their serology. This study aimed to assess factors associated with failure to return for HIV test results. Patients visiting the Fernand-Widal CDAG (Paris) for an HIV test in January-February 2011 were eligible to take part in the study. Data were collected with an anonymous self-administered questionnaire. Factors associated with failure to return were assessed using logistic regression models. Of the 710 participants (participation rate 88%), 46 patients failed to return. Not specifying birthplace and not living in the region of Paris were associated with failure to return. Those who perceived no risk of HIV infection and those who felt they were more at risk than other people were both statistically associated with failure to return. Self-perceived risk seemed to be of chief concern for failure to return for HIV test results and should be considered during pre-test counselling.

Impact of unemployment variations on suicide mortality in Western European countries (2000-2010)

BACKGROUND

A scientific debate is currently taking place on whether the 2008 economic crisis caused an increase in suicide rates. Our main objective was to assess the impact of unemployment rate on suicide rate in Western European countries between 2000 and 2010. We then tried to estimate the excess number of suicides attributable to the increase of unemployment during the 2008-2010 economic crisis.

METHODS

The yearly suicide rates were modelled using a quasi-Poisson model, controlling for sex, age, country and a linear time trend. For each country, the unemployment-suicide association was assessed, and the excess number of suicides attributable to the increase of unemployment was estimated. Sensitivity analyses were performed, notably in order to evaluate whether the unemployment-suicide association found was biased by a confounding context effect ('crisis effect').

RESULTS

A significant 0.3% overall increase in suicide rate for a 10% increase in unemployment rate (95% CI 0.1% to 0.5%) was highlighted. This association was significant in three countries: 0.7% (95% CI 0.0% to 1.4%) in the Netherlands, 1.0% (95% CI 0.2% to 1.8%) in the UK and 1.9% (95% CI 0.8% to 2.9%) in France, with a significant excess number of suicides attributable to unemployment variations between 2008 and 2010 (respectively 57, 456 and 564). The association was modified inconsistently when adding a 'crisis effect' into the model.

CONCLUSIONS

Unemployment and suicide rates are globally statistically associated in the investigated countries. However, this association is weak, and its amplitude and sensitivity to the 'crisis effect' vary across countries. This inconsistency provides arguments against its causal interpretation.

[Mapping of the key oncology indicators available in France]

BACKGROUND

Available data in the field of oncology in France are scattered due to the large number of available indicators and their sources. In order to facilitate identification and analysis of these indicators, the French National Cancer Institute (INCa) has mapped the main indicators available in oncology.

METHODS

Mapping was based on the needs of various categories of potential users. Standardized interviews were conducted face-to-face or by email among representatives to determine their needs and expectations. The underlying data sources were also identified: databases, national surveys, websites. A first selection of indicators was proposed in the report entitled "La situation du cancer en France en 2009" ("The state of cancer in France in 2009") and was expanded. Data collection concerning indicators was performed among INCa correspondents for each theme.

RESULTS

Several themes were defined: epidemiology, prevention and risk factors, screening, medical demography, health care offer, living conditions, costs and expenses, research. Data were classified according to: geographical coverage, age, gender, type of cancer, occupational categories. This information was collected for each indicator selected and was made available via the cancer data website (http://lesdonnees.e-cancer.fr).

CONCLUSIONS

The available oncology indicators are numerous and scattered. Mapping can be a useful tool to facilitate access to these indicators. It should be regularly updated to reflect the most recent data.

Les facteurs associés à la variation de la mortalité par suicide en France

L’objectif de cette communication sera de faire l’état des lieux des connaissances épidémiologiques de la mortalité par suicide, afin de mettre en évidence les disparités sociales et territoriales du suicide en France. Les données disponibles en France sur la mortalité par suicide sont issues de la base de données des causes médicales de décès du CépiDc-Inserm [2], établie à partir des certificats médicaux de décès français. Le CépiDc-Inserm est en charge de la production de ces données, de leur diffusion et de leur analyse [4]. Nous présenterons les indicateurs les plus récents, selon différents axes d’analyses territoriaux et sociaux. L’évolution de ces indicateurs sur la dernière décennie sera discutée. La mortalité par suicide touche trois fois plus les hommes que les femmes en France. Le taux augmente de façon particulièrement forte avec l’âge chez les hommes à partir de 65 ans. La répartition géographique est inégale, le taux étant plus élevé en Bretagne et dans le nord de la France, chez les femmes comme chez les hommes. Une estimation des conséquences imputables à la hausse du chômage pendant la crise économique de 2008 sur le suicide en France sera présentée. La sensibilité de cette estimation a différentes hypothèses, ainsi que sa variabilité selon l’âge, le sexe ou la région de résidence seront discutées. Enfin, les limites des données de mortalité par suicide seront discutées notamment en ce qui concerne la sous déclaration des suicides dans les certificats de décès [1,3].