¹Medizinischer Dienst Nord (MD Nord), Lübeck, Germany
²Kassenärztliche Vereinigung Schleswig-Holstein (KVSH), Bad Segeberg, Germany

Text

Background: Age and polypharmacy, as well as the primary pharmacological treatment at the center of an individual patient’s care, are crucial both for patient-related treatment outcomes and for the economic background of healthcare. For patients, the cooperation of all physicians involved in treatment and the necessary communication processes are essential. The medication plan and the electronic health record could provide far better support than is currently the case. The General Data Protection Regulation (GDPR) restricts patient-related communication between treating physicians, whereas no such limitations exist for anonymized epidemiological analyses. A 10-year analysis allows the identification of trends in healthcare provision that describe multiple aspects of success-oriented cooperation in the historical use of existing drugs, their benefits and limitations – including restricted availability and drug–drug interaction issues – arising from medical prescriptions.

Materials and Methods: All prescriptions issued by physicians in Schleswig-Holstein for patients with statutory health insurance in Germany between 2014 and 2023 were analyzed. The evaluation was conducted across physicians on a per-patient basis. Patient data were pseudonymized. Medications were identified via a pharmaceutical central number, from which the international ATC code with German specification was derived from the corresponding database. For each patient and year, drug costs could be aggregated at the ATC four-character level (third level). The ATC-4 code with the highest cost determined the Morbidity Related Group (MRG). The number of ATC-4 codes per patient served as a cross-physician measure of polypharmacy. For each MRG and year, averages for age and polypharmacy across all patients were calculated, along with the number of affected patients. For each MRG, linear trends in age, polypharmacy, and patient numbers were determined over the observation period. Correlations and partial correlations of these three variables were then calculated as a function of the MRG. In addition to linear trends, changes between specific years were examined. Pairwise comparisons of 2019, 2021, and 2023 enabled analysis of the parameters and their correlations/partial correlations before, during, and after the COVID-19 pandemic. The modulus value derived from correlations indicates whether all partial correlations are numerically greater or smaller than the corresponding correlations. Correlations and partial correlations can be translated into relationships in spherical trigonometry and Jacobian elliptic functions.

Results: There are MRGs in which the variables age and polypharmacy display essentially linear increasing or decreasing trends; a threshold of absolute deviations from the linear trend can be used to identify such cases. Of further interest are MRGs in which a trend reversal occurs within the 10-year period.

Across all MRGs with at least 100 patients, correlations between patient numbers, mean age, and mean polypharmacy were found to be 0.216, 0.663, and 0.405, respectively, with a modulus value of 1.044. Corresponding partial correlations were –0.073, 0.644, and 0.356.

Due to the modulus value of 1.044 being slightly above 1, all partial correlations were smaller than the correlations, with a sign reversal in one case. The modulus value was close to the critical threshold of 1.

To assess the impact of the COVID-19 pandemic on the number of patients with prescriptions, correlations were analyzed between 2019–2021, 2019–2023, and 2021–2023, representing changes before, during, and after the pandemic. The resulting values were 0.985, 0.988, and 0.978. Here, a comparatively high modulus value of 21.999 was obtained, with partial correlations of 0.581, 0.675, and 0.195.

In this case, the modulus value of about 22 is substantially larger than the critical threshold of 1, and no sign reversal of partial correlations occurred. The correlations for COVID-related changes were close to 1. Excluding one of the three variables in each case led to a markedly reduced partial correlation between the remaining two variables.

Conclusion: There are MRGs in which the variables age and polypharmacy display essentially linear increasing or decreasing trends; a threshold of absolute deviations from the linear trend can be used to identify such cases. Of further interest are MRGs in which a trend reversal occurs within the 10-year period.

Across all MRGs with at least 100 patients, correlations between patient numbers, mean age, and mean polypharmacy were found to be 0.216, 0.663, and 0.405, respectively, with a modulus value of 1.044. Corresponding partial correlations were –0.073, 0.644, and 0.356.

Due to the modulus value of 1.044 being slightly above 1, all partial correlations were smaller than the correlations, with a sign reversal in one case. The modulus value was close to the critical threshold of 1.

To assess the impact of the COVID-19 pandemic on the number of patients with prescriptions, correlations were analyzed between 2019–2021, 2019–2023, and 2021–2023, representing changes before, during, and after the pandemic. The resulting values were 0.985, 0.988, and 0.978. Here, a comparatively high modulus value of 21.999 was obtained, with partial correlations of 0.581, 0.675, and 0.195.

In this case, the modulus value of about 22 is substantially larger than the critical threshold of 1, and no sign reversal of partial correlations occurred. The correlations for COVID-related changes were close to 1. Excluding one of the three variables in each case led to a markedly reduced partial correlation between the remaining two variables.

Literatur

[1] Armitage JV, Eberlein WF. Elliptic functions. Cambridge University Press; 2006. (London Mathematical Society Student Texts; 67).
[2] Brief AP, Burke MJ, George JM, Robinson BS, Webster J. Should negative affectivity remain an unmeasured variable in the study of job stress? J Appl Psychol. 1988 May;73(2):193-8. doi: 10.1037/0021-9010.73.2.193
[3] Good I. C389. Partial correlation and spherical trigonometry, II. Journal of Statistical Computation and Simulation. 1992;43(3-4):253. DOI: 10.1080/00949659208811442
[4] Janssen J, Laatz W. Korrelation. Statistische Datenanalyse mit SPSS für Windows: Eine anwendungsorientierte Einführung in das Basissystem. 1994. p. 347-357.
[5] Johnell K, Klarin I. The relationship between number of drugs and potential drug-drug interactions in the elderly: a study of over 600,000 elderly patients from the Swedish Prescribed Drug Register. Drug Saf. 2007;30(10):911-8. DOI: 10.2165/00002018-200730100-00009
[6] Kim S. ppcor: an r package for a fast calculation to semi-partial correlation coefficients. Communications for statistical applications and methods. 2015;22(6):665.
[7] Runkler TA. Datenanalyse und Modellierung. Information Mining: Methoden, Algorithmen und Anwendungen intelligenter Datenanalyse. Vieweg; 2000. p. 53-109.
[8] Schuster R, Emcke T, von Arnstedt E, Heidbreder M. Morbidity related groups (mrg) for epidemiological analysis in outpatient treatment. In: Exploring complexity in health: An interdisciplinary systems approach. IOS Press; 2016. (Studies in Health Technology and Informatics; 228). p. 783-787. DOI: 10.3233/978-1-61499-678-1-783
[9] Schuster R, Ostermann T, Heidbreder M, Emcke T. Relations of Morbidity Related Groups (MRG), ICD-10 Codes and Age and Gender Structure in Outpatient Treatment. In: Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2018) - Volume 5: HEALTHINF; 2018 Jan 19-21; Funchal, Madeira, Portugal. SCITEPRESS; 2018. p. 322-328. DOI: 10.5220/0006544303220328