Data from national health registers as endpoints for the Tromsø Study: Correctness and completeness of stroke diagnoses

Consistency between the National Patient Register and the Swedish Cancer Register

PURPOSE

The Swedish National Patient Register (NPR) is widely used as a data source in epidemiological studies, but the consistency of all cancer diagnoses compared to the Swedish Cancer Register (SCR) remains unclear. Using NPR supplementary for detecting safety signals is beneficial due to shorter data extraction delays compared to using SCR alone. This study aims to evaluate the consistency of NPR for cancer diagnoses compared to SCR and its potential use in pharmacoepidemiology.

METHODS

Patients with a cancer diagnosis recorded in SCR during 2018-2020 were included. To measure the consistency of NPR diagnoses with SCR as the gold standard, positive predictive value (PPV), and sensitivity were calculated. As an empirical example showing differences in identification of cancer diagnoses in NPR and SCR, two nested case-control studies for the association between antidiabetic medications and pancreatic cancer were repeated using the two registers. Conditional logistic regression was performed and the 95% confidence intervals (CIs) for the odds ratios (ORs) were checked for overlaps.

RESULTS

For breast, male genital organs, and oral cancers consistency was high (PPV: 87.5%-97.4%, sensitivity: 82.2%-91.0%), while for female genital organs, thyroid, and ill-defined, secondary, and unspecified sites cancers it was low (PPV: 8.8%-90.0%, sensitivity: 19.9%-32.3%). All the CIs for the ORs from the nested case-control studies overlapped when pancreatic cancer was identified in NPR or SCR.

CONCLUSION

Consistency of cancer diagnoses in NPR when compared to SCR depends on cancer type with higher consistency for some cancers and lower for others. Differences in diagnostic processes for different cancer types and coding of cancer in the two registers may explain part of the inconsistent results.

Prognostic significance of chronic myocardial injury diagnosed by three different cardiac troponin assays in patients admitted with suspected acute coronary syndrome

OBJECTIVES

Chronic myocardial injury (CMI) is defined as stable concentrations of cardiac troponin T or I (cTnT or cTnI) above the assay-specific 99th percentile upper reference limit (URL) and signals poor outcome. The clinical implications of diagnosing CMI are unclear. We aimed to assess prevalence and association of CMI with long-term prognosis using three different high-sensitivity cTn (hs-cTn) assays.

METHODS

A total of 1,292 hospitalized patients without acute myocardial injury had cTn concentrations quantified by hs-cTn assays by Roche Diagnostics, Abbott Diagnostics and Siemens Healthineers. The median follow-up time was 4.1 years. The prevalence of CMI and hazard ratios for mortality and cardiovascular (CV) events were calculated based on the URL provided by the manufacturers and compared to the prognostic accuracy when lower percentiles of cTn (97.5, 95 or 90), limit of detection or the estimated bioequivalent concentrations between assays were used as cutoff values.

RESULTS

There was no major difference in prognostic accuracy between cTnT and cTnI analyzed as continuous variables. The correlation between cTnT and cTnI was high (r=0.724-0.785), but the cTnT assay diagnosed 3.9-4.5 times more patients with having CMI based on the sex-specific URLs (TnT, n=207; TnI Abbott, n=46, TnI Siemens, n=53) and had higher clinical sensitivity and AUC at the URL.

CONCLUSIONS

The prevalence of CMI is highly assay-dependent. cTnT and cTnI have similar prognostic accuracy for mortality or CV events when measured as continuous variables. However, a CMI diagnosis according to cTnT has higher prognostic accuracy compared to a CMI diagnosis according to cTnI.

Thrombolytic Treatment in Wake-Up Stroke: A Propensity Score-Matched Analysis of Treatment Effectiveness in the Norwegian Stroke Registry

BACKGROUND

Previous clinical trials found improved outcome of thrombolytic treatment in patients with ischemic wake-up stroke (WUS) selected by advanced imaging techniques. The authors assessed the effectiveness of thrombolytic treatment in patients with WUS in a nationwide stroke registry.

METHODS AND RESULTS

Using propensity score matching, the authors assessed the effectiveness and safety of thrombolytic treatment versus no thrombolytic treatment in 726 patients (363 matched pairs) with WUS in the Norwegian Stroke Registry in 2014 to 2019. Thrombolytic treatment in WUS versus known-onset stroke was compared in 730 patients (365 matched pairs). Functional outcomes were assessed by the modified Rankin Scale (mRS) at 3 months. A significant benefit of thrombolytic treatment in WUS was seen in ordinal analysis (odds ratio [OR], 1.48 [95% CI, 1.15-1.91]; P=0.003) and for mRS 0 to 2 (OR, 1.81 [95% CI, 1.29-2.52]; P=0.001) but not for mRS 0 or 1 (OR, 1.32 [95% CI, 1.00-1.74]; P=0.050). The proportion of patients with mRS 0 or 1 was lower in patients with WUS who underwent thrombolysis versus those with known-onset stroke (50.4% versus 59.5%; OR, 0.69 [95% CI, 0.52-0.93]; P=0.013), while outcomes were similar between groups for mRS 0 to 2 and ordinal analysis. Symptomatic intracranial hemorrhage after thrombolytic treatment occurred in 4.4% of patients with WUS and 3.9% of patients with known-onset stroke (OR, 1.14 [95% CI, 0.54-2.41]; P=0.726).

CONCLUSIONS

Thrombolytic treatment in patients with WUS was associated with improved functional outcome compared with patients with no thrombolytic treatment and was not associated with increased rates of symptomatic intracranial hemorrhage compared with known-onset stroke. The results indicate that thrombolytic treatment is effective and safe in WUS in a real-life setting.

Pain tolerance after stroke: The Tromsø study

BACKGROUND

Stroke lesions might alter pain processing and modulation by affecting the widely distributed network of brain regions involved. We aimed to compare pain tolerance in stroke survivors and stroke-free persons in the general population, with and without chronic pain.

METHODS

We included all participants of the sixth and seventh wave of the population-based Tromsø Study who had been tested with the cold pressor test (hand in cold water bath, 3°C, maximum time 106 s in the sixth wave and 120 s in the seventh) and who had information on previous stroke status and covariates. Data on stroke status were obtained from the Tromsø Study Cardiovascular Disease Register and the Norwegian Stroke Register. Cox regression models were fitted using stroke prior to study attendance as the independent variable, cold pressor endurance time as time variable and hand withdrawal from cold water as event. Statistical adjustments were made for age, sex, diabetes, hypertension, hyperlipidaemia, body mass index and smoking.

RESULTS

In total 21,837 participants were included, 311 of them with previous stroke. Stroke was associated with decreased cold pain tolerance time, with 28% increased hazard of hand withdrawal (hazard ratio [HR] 1.28, 95% CI 1.10-1.50). The effect was similar in participants with (HR 1.28, 95% CI 0.99-1.66) and without chronic pain (HR 1.29, 95% CI 1.04-1.59).

CONCLUSIONS

Stroke survivors, with and without chronic pain, had lower cold pressor pain tolerance, with possible clinical implications for pain in this group.

SIGNIFICANCE

We found lower pain tolerance in participants with previous stroke compared to stroke-free participants of a large, population-based study. The association was present both in those with and without chronic pain. The results may warrant increased awareness by health professionals towards pain experienced by stroke patients in response to injuries, diseases and procedures.

Sex-specific time trends in incident atrial fibrillation and the contribution of risk factors: the Tromsø Study 1994-2016

AIMS

To explore sex-specific time trends in atrial fibrillation (AF) incidence and to estimate the impact of changes in risk factor levels using individual participant-level data from the population-based Tromsø Study 1994-2016.

METHODS AND RESULTS

A total of 14 818 women and 13 225 men aged 25 years or older without AF were enrolled in the Tromsø Study between 1994 and 2008 and followed up for incident AF throughout 2016. Poisson regression was used for statistical analyses. During follow-up, age-adjusted AF incidence rates in women decreased from 1.19 to 0.71 per 1000 person-years. In men, AF incidence increased from 1.18 to 2.82 per 1000 person-years in 2004, and then declined to 1.94 per 1000 person-years in 2016. Changes in systolic blood pressure (SBP) and diastolic blood pressure (DBP), body mass index (BMI), physical activity, smoking and alcohol consumption together accounted for 10.9% [95% confidence interval (CI): -2.4 to 28.6] of the AF incidence decline in women and for 44.7% (95% CI: 19.2; 100.0) of the AF incidence increase in men. Reduction in SBP and DBP had the largest contribution to the decrease in AF incidence in women. Increase in BMI had the largest contribution to the increase in AF incidence in men.

CONCLUSION

In the population-based Tromsø Study 1994-2016, AF incidence decreased in women and increased following a reverse U-shape in men. Individual changes in SBP and DBP in women and individual changes in BMI in men were the most important risk factors contributing to the AF incidence trends.

Wake-up stroke and unknown-onset stroke; occurrence and characteristics from the nationwide Norwegian Stroke Register

Introduction

Population-based knowledge of the characteristics of wake-up stroke and unknown-onset stroke is limited. We compared occurrence and characteristics of ischaemic and haemorrhagic wake-up stroke, unknown-onset stroke and known-onset stroke in a nationwide register-based study.

Patients and methods

We included patients registered in the Norwegian Stroke Register from 2012 through 2019. Age, sex, risk factors, clinical characteristics, acute stroke treatment and discharge destination were compared according to stroke type and time of onset.

Results

Of the 60,320 patients included, 11,451 (19%) had wake-up stroke, 11,098 (18.4%) had unknown time of onset and 37,771 (62.6%) had known symptom onset. The proportion of haemorrhagic stroke was lower among wake-up stroke patients (1107/11,451, 9.7%, 95% CI: 9.1-10.2) than for known-onset stroke (5230/37,771, 13.8%, 95% CI: 13.5-14.2) and for unknown-onset stroke (1850/11,098, 16.7%, 95% CI: 16.0-17.4). Mild stroke (NIHSS <5) was more frequent in ischaemic wake-up stroke (5364/8308, 64.6%, 95% CI: 63.5-65.5) than in known-onset (16,417/26,746, 61.4%, 95% CI: 60.8-62.0) and unknown-onset stroke (3242/5853, 55.4%, 95% CI: 54.1-56.7), while baseline characteristics were otherwise similar to known-onset stroke. Unknown-onset stroke patients were more often female, lived alone and had more severe strokes compared to wake-up stroke and known-onset stroke patients. Unknown-onset stroke patients were more often in need of community-based health care on discharge and had a higher in-hospital mortality.

Discussion and conclusions

Ischaemic wake-up strokes shared baseline characteristics with known-onset strokes, but tended to be milder. Ischaemic unknown-onset stroke patients differed significantly from wake-up stroke, emphasising the importance of considering them as separate entities.