Using statutory health insurance data to evaluate non-response in a cross-sectional study on depression among patients with diabetes in Germany

Prospective health care costs and lost work days associated with diabetes-related distress and depression symptoms among 1488 individuals with diabetes

The aim of this study was to investigating the impact of major depression symptoms and diabetes-related distress on future health care costs and lost workdays in individuals with diabetes. We linked survey data from a random sample of a German statutory health insurance (SHI) with diabetes (n = 1488, 63.0% male, mean age 66.9 years) with their SHI data one year after the survey. Within the survey data we identified major depression symptoms (Patient Health Questionnaire-9) and diabetes-related distress (Problem Areas in Diabetes Scale). We retrieved health care costs and lost workdays from SHI data. To assess the impact of major depression symptoms and diabetes-related distress on health care costs and lost workdays, we adjusted regression models for age, sex, education, employment status, and diabetes duration, type, and severity. Major depression symptoms were associated with significantly higher costs (by a factor of 1.49; 95% CI: 1.18-1.88). Lost workdays were also more likely for respondents with depression symptoms (RR1.34; 0.97-1.86). Health care costs (by a factor of 0.81; 0.66-1.01) and the risk of lost workdays (RR 0.86; 0.62-1.18) may be lower among respondents with high diabetes-related distress. While major depression and diabetes-related distress have overlapping indicators, our results indicate different impacts on health care costs.

Linking health survey data with health insurance data: methodology, challenges, opportunities and recommendations for public health research. An experience from the HISlink project in Belgium

In recent years, the linkage of survey data to health administrative data has increased. This offers new opportunities for research into the use of health services and public health. Building on the HISlink use case, the linkage of Belgian Health Interview Survey (BHIS) data and Belgian Compulsory Health Insurance (BCHI) data, this paper provides an overview of the practical implementation of linking data, the outcomes in terms of a linked dataset and of the studies conducted as well as the lessons learned and recommendations for future links.Individual BHIS 2013 and 2018 data was linked to BCHI data using the national register number. The overall linkage rate was 92.3% and 94.2% for HISlink 2013 and HISlink 2018, respectively. Linked BHIS-BCHI data were used in validation studies (e.g. self-reported breast cancer screening; chronic diseases, polypharmacy), in policy-driven research (e.g., mediation effect of health literacy in the relationship between socioeconomic status and health related outcomes, and in longitudinal study (e.g. identifying predictors of nursing home admission among older BHIS participants). The linkage of both data sources combines their strengths but does not overcome all weaknesses.The availability of a national register number was an asset for HISlink. Policy-makers and researchers must take initiatives to find a better balance between the right to privacy of respondents and society's right to evidence-based information to improve health. Researchers should be aware that the procedures necessary to implement a link may have an impact on the timeliness of their research. Although some aspects of HISlink are specific to the Belgian context, we believe that some lessons learned are useful in an international context, especially for other European Union member states that collect similar data.

Exploring potential for selection bias in using survey data to estimate the association between institutional trust and depression

PURPOSE

We tested the hypothesis that low institutional trust would be associated with depressive symptom elevation, with attention to potential selection bias.

METHODS

The District of Columbia Area Survey (DCAS) was conducted by mail in 2018. Invitations sent to 8800 households resulted in a sample of 1061 adults. Institutional trust questions referenced nonprofit organizations, businesses, and government. Depressive symptom elevation was assessed using PHQ-9. Logistic regression model estimates were compared with and without adjustment for sociodemographic characteristics and neighborhood satisfaction; among complete cases and following multiple imputation of missing covariate data; and with and without survey weights or correction for collider selection bias.

RESULTS

Of 968 participants without missing depressive symptom or trust data, 24% reported low institutional trust. Low institutional trust was associated with elevated depressive symptoms (adjusted OR following multiple imputation: 2.0; 95% CI: 1.1, 3.4), although the association was attenuated with use of survey weights (adjusted OR incorporating multiple imputation and survey weights: 1.6; 95% CI: 0.7, 3.2).

CONCLUSIONS

Under contrasting scenarios where low institutional trust and depressive symptoms jointly increase nonresponse, selection bias could lead to under- or overestimation of this association. Future research could explore posited selection bias scenarios that differ in direction of bias.

Diabetes-related research priorities of people with type 1 and type 2 diabetes: a cross-sectional study in Germany

To investigate (i) the importance and priorities of research objectives for people with type 1 (T1DM) and type 2 diabetes (T2DM); (ii) subgroups with specific research priorities; (iii) associated factors (e.g., sociodemographic characteristics) of the subgroups. The cross-sectional survey was conducted in 2018 using data from 869 respondents (29.0% response, 31.2% female, mean age 61.3 years, 62.7% T2DM) from a German statutory health insurance population. Diabetes-related research priorities were assessed with a questionnaire. Subgroups and associated factors were identified using latent class analysis. Three subgroups were found in T1DM: (1) high priority for the research topic 'healing diabetes' and moderate priority for the research topic 'prevention of long-term complications', (2) priorities for simplifying handling (high) and stress reduction (moderate), (3) priorities for healing diabetes (high) and simplifying handling (high). Three subgroups were found in T2DM: (1) priorities for simplifying handling (moderate), diabetes prevention (moderate) and prevention of long-term complications (moderate), (2) priorities for stress reduction (high) and diabetes prevention (moderate), (3) priorities for simplifying handling (high) and stress reduction (high). Classes differed in age and HbA_1c. Knowledge about research priorities enables researchers to align their work with the needs of people with diabetes.

What Can Legacy Patient-Reported Outcome Measures Tell Us About Participation Bias in Patient-Reported Outcomes Measurement Information System Scores Among Lumbar Spine Patients?

Objective

Patient-Reported Outcomes Measurement Information System (PROMIS) is a validated tool for assessing patient reported outcomes in spine surgery. However, PROMIS is vulnerable to non-response bias. The purpose of this study is to characterize differences in patient reported outcome measure (PROM) scores between patients who do and do not complete PROMIS physical function (PF) surveys following lumbar spine surgery.

Methods

A prospectively maintained database was retrospectively reviewed for primary, elective lumbar spine procedures from 2015 to 2019. Outcome measures for Patient Health Questionnaire (PHQ-9), Visual Analogue Score (VAS) back & leg, Oswestry Disability Index (ODI), and 12-Item Short Form Physical Composite Summary (SF-12 PCS) were recorded at both preoperative and postoperative (6-week, 12-week, 6-month, 1-year, 2-year) timepoints. Completion rates for PROMIS PF surveys were recorded and patients were categorized into groups based on completion. Differences in mean scores at each timepoint between groups was determined.

Results

809 patients were included with an average age of 48.1 years. No significant differences were observed for all outcome measures between PROMIS completion groups preoperatively. Postoperative PHQ-9, VAS back, VAS leg, and ODI scores differed significantly between groups through 1-year (all p<0.05). SF-12 PCS differed significantly only at 6-weeks (p=0.003).

Conclusion

Patients who did not complete PROMIS PF surveys had significantly poorer outcomes than those that did in terms of postoperative depressive symptoms, pain, and disability. This suggests that patients completing PROMIS questionnaires may represent a healthier cohort than the overall lumbar spine population.

Insulin Pump Treatment in Adults with Type 1 Diabetes in the Capital Region of Denmark: Design and Cohort Characteristics of the Steno Tech Survey

INTRODUCTION

Insulin pump therapy can improve quality of life and glycaemic outcomes for many people with type 1 diabetes (T1D). The multidimensional Steno Tech Survey study aims to investigate why some insulin pump users do not achieve treatment goals. In this article, we present the study design and analyse differences in population characteristics between responders and non-responders.

METHODS

In June 2020, all 1591 insulin pump users (≥ 18 years) in the Capital Region of Denmark were invited to participate in an online questionnaire that evaluated several dimensions of insulin pump self-management and psychosocial health. Demographic, socioeconomic and clinical characteristics, including age, sex and HbA1c, of the cohort were identified via national registries. Predictors of questionnaire response/non-response were explored with logistic regression analysis.

RESULTS

In the full study population, 58% were female, median age was 42 years and median HbA1c was 58 mmol/mol (7.5%); 30% had HbA1c < 53 mmol/mol (7.0%). In total, 770 individuals (48%) responded to the questionnaire. Logistic regression analysis showed that 50+ years of age (odds ratio [OR] = 2.3, 95% confidence interval [CI]: 1.4-3.8), female sex (OR = 1.3, CI: 1.02-1.6), being married (OR = 1.8, CI: 1.3-2.4) and having long higher education (OR = 1.6, CI: 1.004-2.5) were significantly associated with a higher likelihood of responding to the survey; the opposite was found for HbA1c from 64 to < 75 mmol (8.0-9.0%) (OR = 0.6, CI: 0.4-0.8) and HbA1c ≥ 75 mmol/mol (≥ 9.0%) (OR = 0.2, CI: 0.1-0.3).

CONCLUSIONS

The established Steno Tech cohort enables future analysis of a range of psychosocial and behavioural aspects of insulin pump self-management. Interpretation and generalization of findings should consider observed differences between responders and non-responders.

Health Care Use and Costs in Individuals With Diabetes With and Without Comorbid Depression in Germany: Results of the Cross-sectional DiaDec Study

OBJECTIVE

Increased health care use and costs have been reported in individuals with diabetes with comorbid depression. Knowledge regarding cost differences between individuals with diabetes alone and those with diabetes and diagnosed/undiagnosed depression is, however, scarce. We therefore compared use and costs for patients with diabetes and no depression and patients with diabetes and documented depression diagnosis or self-reported depression symptoms for several cost components, including mental health care costs.

RESEARCH DESIGN AND METHODS

Data from a 2013 cross-sectional survey of randomly sampled members of a nationwide German statutory health insurance (SHI) provider with diabetes (n = 1,634) were linked individually with SHI data covering four quarters before and after the survey. Self-reported depression symptoms were assessed with the Patient Health Questionnaire-9, with depression diagnosis taken from SHI data. We analyzed health care use and costs, using regression analysis to calculate cost ratios (CRs) with adjustment for sociodemographic/socioeconomic factors and comorbidities for two groups: 1) those with no symptoms and no diagnosis and 2) those with symptoms or diagnosis. In our explorative subanalysis we analyzed subgroups with either symptoms or diagnosis separately.

RESULTS

Annual mean total health care costs were higher for patients with comorbid depression (EUR 5,629 [95% CI 4,987-6,407]) than without (EUR 3,252 [2,976-3,675], the CR being 1.25 [1.14-1.36]). Regression analysis showed that excess costs were highly associated with comorbidities. Mental health care costs were very low for patients without depression (psychotherapy EUR 2; antidepressants EUR 4) and still relatively low for those with depression (psychotherapy EUR 111; antidepressants EUR 76).

CONCLUSIONS

Costs were significantly higher when comorbid depression was present either as symptoms or diagnosed. Excess costs for mental health services were rather low.

Factors Associated with Survey Non-Response in a Cross-Sectional Survey of Persons with an Axial Spondyloarthritis or Osteoarthritis Claims Diagnosis

Non-response in surveys can lead to bias, which is often difficult to investigate. The aim of this analysis was to compare factors available from claims data associated with survey non-response and to compare them among two samples. A stratified sample of 4471 persons with a diagnosis of axial spondyloarthritis (axSpA) and a sample of 8995 persons with an osteoarthritis (OA) diagnosis from a German statutory health insurance were randomly selected and sent a postal survey. The association of age, sex, medical prescriptions, specialist physician contact, influenza vaccination, hospitalization, and Elixhauser comorbidity index with the survey response was assessed. Multiple logistic regression models were used with response as the outcome. A total of 47% of the axSpA sample and 40% of the OA sample responded to the survey. In both samples, the response was highest in the 70-79-year-olds. Women in all age groups responded more often, except for the 70-79-year-olds. Rheumatologist/orthopedist contact, physical therapy prescription, and influenza vaccination were more frequent among responders. In the logistic regression models, rheumatologist/orthopedist treatment, influenza vaccination, and physical therapy were associated with a higher odds ratio for response in both samples. The prescription of biologic drugs was associated with higher response in axSpA. A high Elixhauser comorbidity index and opioid use were not relevantly associated with response. Being reimbursed for long-term care was associated with lower response-this was only significant in the OA sample. The number of quarters with a diagnosis in the survey year was associated with higher response. Similar factors were associated with non-response in the two samples. The results can help other investigators to plan sample sizes of their surveys in similar settings.