Probabilistic Quantification of Bias to Combine the Strengths of Population-Based Register Data and Clinical Cohorts-Studying Mortality in Osteoarthritis

Use of quantitative bias analysis to evaluate single-arm trials with real-world data external controls

PURPOSE

Use of real-world data (RWD) for external controls added to single-arm trials (SAT) is increasingly prevalent in regulatory submissions. Due to inherent differences in the data-generating mechanisms, biases can arise. This paper aims to illustrate how to use quantitative bias analysis (QBA).

METHODS

Advanced non-small cell lung cancer (NSCLC) serves as an example, where many small subsets of patients with molecular tumor subtypes exist. First, some sources of bias that may occur in oncology when comparing RWD to SAT are described. Second, using a hypothetical immunotherapy agent, a dataset is simulated based on expert input for survival analysis of advanced NSCLC. Finally, we illustrate the impact of three biases: missing confounder, misclassification of exposure, and outcome evaluation.

RESULTS

For each simulated scenario, bias was induced by removing or adding data; hazard ratios (HRs) were estimated applying conventional analyses. Estimating the bias-adjusted treatment effect and uncertainty required carefully selecting the bias model and bias factors. Although the magnitude of each biased and bias-adjusted HR appeared moderate in all three hypothetical scenarios, the direction of bias was variable.

CONCLUSION

These findings suggest that QBA can provide an intuitive framework for bias analysis, providing a key means of challenging assumptions about the evidence. However, the accuracy of bias analysis is itself dependent on correct specification of the bias model and bias factors. Ultimately, study design should reduce bias, but QBA allows us to evaluate the impact of unavoidable bias to assess the quality of the evidence.

Determining occupational accidents baseline ratios by considering a synthetic population: The case of Spain

In most countries, a government agency or collaborating organization gathers information on occupational accidents. Comparisons based on a single factor such as autonomous community, activity sector or others, often leads to contradictory conclusions. The use of this information for comparison is not immediate because the different characteristics considered give place to different possible comparisons. The elaboration of a single baseline for each set of characteristics is addressed. The method proposed comes from the data available in Spain but could be applied to other cases. The method consists of: (1) selecting factors-those selected are age, sex, autonomous community and activity; (2) the generation of a synthetic population based on data from a survey and general proportions by applying the Optimal Representative Sample Weighting (rsw); and (3) the prediction of the accidents ratio for each set of characteristic by using a XGBoost decision trees ensemble. The results confirm the appropriateness of the method.

Knee and hip osteoarthritis and risk of nine cancers in a large real-world matched cohort study

OBJECTIVES

Joint replacement due to end-stage osteoarthritis (OA) has been linked to incidence of several cancers. We aimed to estimate the association between newly diagnosed knee and hip OA and incidence of nine common cancer types.

METHODS

We identified persons with incident knee or hip OA, aged ≥40 years, between 2009 and 2015 in the SIDIAP database in Catalonia, Spain. We matched up to 3 OA-free controls on age, sex and general practitioner. We followed participants from 1 year after OA diagnosis until migration, death, end of study at Dec 31st 2017 or incident cancer of: stomach, colorectal, liver, pancreas, lung, skin, breast, prostate, and bladder. We used flexible parametric survival models, adjusted for confounders. Estimates were corrected for misclassification using probabilistic bias analysis.

RESULTS

We included 117 750 persons with knee OA and matched 309 913 persons without, with mean (SD) age of 67.5 (11.1) years and 63% women. The hip cohort consisted of 39 133 persons with hip OA and 116 713 controls. For most of included cancers, the hazard ratios (HRs) were close to 1. The HR of lung cancer for knee OA exposure was 0.80 (95%CI 0.71, 0.89) and attenuated to 0.98 (0.76, 1.27) in non-smokers. The hazard of colorectal cancer was lower in persons with both knee and hip OA by 10-20%.

CONCLUSIONS

Knee and hip OA are not associated with studied incident cancers, apart from lower risk of colorectal cancer. The often-reported protective association of knee OA with lung cancer is explained by residual confounding.

Socioeconomic inequalities in all-cause and cause-specific mortality among patients with osteoarthritis in the Skåne region, Sweden

OBJECTIVE

To assess the association between education and all-cause and cause-specific mortality among patients with osteoarthritis (OA) in comparison to an OA-free reference cohort.

METHODS

Using data from the Skåne healthcare register, we identified all residents aged ≥45 years in the region of Skåne, with doctor-diagnosed OA of peripheral joints between 1998 and 2013 (n=123,993). We created an age and sex-matched reference cohort without OA diagnosis (n=121,318). Subjects were followed until death, relocation outside Skåne, or the end of 2014. The relative index of inequality (RII) and the slope index of inequality (SII) were estimated by the Cox model and Aalen´s additive hazard model, respectively.

RESULTS

We found an inverse association between education and mortality. The magnitude of relative inequalities in all-cause mortality were comparable in the OA (RII 1.53, 95% CI:1.46, 1.61) and reference cohorts (RII:1.54, 95% CI:1.47, 1.62). The absolute inequalities were smaller in the OA (SII 937 all-cause deaths per 100,000 person-years, 95% CI:811, 1063) compared with the reference cohort (SII 1265, 95% CI:1109, 1421). Cardiovascular mortality contributed more to the absolute inequalities in the OA than in the reference cohort (60.1% vs. 48.1%) while the opposite was observed for cancer mortality (8.5% vs. 22.3%).

CONCLUSION

We found higher all-cause and cause-specific mortality in OA patients with lower education. The observed inequalities in the OA cohort reflect the inequalities in the population at large. The greater burden of cardiovascular diseases in OA patients suggests that proper management of cardiovascular risk factors in OA patients is important.

Trajectory of excess healthcare consultations, medication use, and work disability in newly diagnosed knee osteoarthritis: a matched longitudinal register-based study

PURPOSE

To estimate the excess healthcare use and work disability attributable to knee osteoarthritis (OA) in the first 5 years following diagnosis.

METHODS

Among individual aged 40-80 years who resided in Skåne on 31^st December 2008, we identified those with a main diagnosis of knee OA during 2009-2014 and no previous diagnosis of any OA from 1998 (n = 16,888). We created a comparison cohort matched (1:1) by sex, age, and municipality from individuals with no OA diagnosis (at any site) during 1998-2016. We compared healthcare use and net disability days for 60 months following diagnosis between the two groups. We applied a survival-adjusted regression technique controlling for sociodemographic characteristics as well as pre-diagnosis outcome and comorbidity.

RESULTS

The estimated 5-year incremental effects of knee OA per-patient were 16.8 (95% CI: 15.8, 17.7) healthcare consultations, 0.7 (0.4, 1.1) inpatient days, 420 (372, 490) defined daily dose of prescribed medications, and 21.8 (15.2, 30.0) net disability days. Primary care consultations constituted about 73% of the excess healthcare consultations. Most of these incremental effects occurred in the first year after diagnosis. Better survival in the knee OA group accounted for 0.7 (95% CI: 0.5, 0.8) and 1.4 (0.7, 2.6) of the excess healthcare consultations and net disability days, respectively. Both estimated total and incremental resources use were generally greater for women than men with knee OA.

CONCLUSION

Knee OA was associated with considerable excess healthcare use and work disability independent of pre-diagnosis resources use, comorbidity, and sociodemographic characteristics.