Adjunct Therapies Tied to Higher Odds of Opioid Discontinuation in German Primary Care

A large analysis of German health-insurance claims has identified which patients prescribed opioids by their general practitioners are most and least likely to have no opioid prescription at the end of the year. Among 94,125 patients who received at least one oral or transdermal opioid prescription from a general practitioner in the first three quarters of 2021, 45.5 percent had no opioid prescription in the fourth quarter, a measure the researchers used as an operational definition of opioid discontinuation. The study, published in Drugs – Real World Outcomes, drew on claims data from AOK Baden-Württemberg, a regional statutory health insurer covering roughly 2.5 million insured individuals.

The strongest positive association involved adjunct therapies, predominantly physiotherapy- and exercise-based interventions along with passive modalities such as massage and thermal applications. Patients who received prescriptions for these non-pharmacological treatments had substantially higher odds of the absence of an opioid prescription in the fourth quarter, with an odds ratio of 1.834, corresponding to an 83.4 percent increase in odds. By contrast, several clinical characteristics were associated with lower odds of discontinuation. A documented somatoform pain disorder was associated with 19.9 percent lower odds (odds ratio 0.801), a substance use disorder with 18.6 percent lower odds (odds ratio 0.814), and a severe depressive episode with 9.7 percent lower odds (odds ratio 0.903), all statistically significant at p < 0.001.

The study was an observational claims-data analysis, and the authors are explicit that the findings should be interpreted as associations rather than causal effects. Claims data record prescriptions, which indicate treatment intent, but cannot confirm actual medication intake, adherence, dose changes, or patient-initiated discontinuation. The outcome measure itself has limitations: the absence of a prescription in a single quarter may misclassify discontinuation, since temporary prescribing gaps, care transitions, or prescriptions outside the captured setting could all produce the same pattern. The authors also note that reasons for tapering or continuing therapy, such as clinical decision-making, patient preference, adverse effects, or worsening pain, are not available in claims records.

The cohort characteristics shaped how the results should be read. Patients were aged 18 or older, with a mean age of 65.5 years, and 62 percent were female. A chronic non-cancer pain diagnosis was documented in 49.3 percent of patients, with “other chronic pain” (ICD-10-GM R52.2) the most frequently coded diagnosis. Patients receiving palliative care or diagnosed with malignant disease were excluded, as were those whose opioid prescriptions came from specialist physicians rather than general practitioners and those with coded opioid-related adverse effects or poisonings during the first three quarters. Weak, long-acting opioids were the most commonly prescribed medications, and long-acting opioids predominated overall, prescribed to 94.4 percent of the cohort.

The prescribing patterns across 2021 showed a distinctive shape. Among all patients with at least one opioid prescription, 31.9 percent had prescriptions in only one quarter, 19.0 percent in two quarters, 13.3 percent in three quarters, and 35.7 percent in all four quarters. The authors describe this as a modest U-shaped pattern suggesting heterogeneous care trajectories: a larger group with short-term opioid exposure and a second group with persistent, year-long prescribing. Strong opioid users were particularly likely to persist, with 52.1 percent receiving prescriptions in all four quarters, compared with 26.4 percent of weak-opioid recipients. Tilidine was the most frequently prescribed weak opioid, reaching 60,898 patients, while oxycodone was the most frequently prescribed strong opioid, prescribed to 10,744 patients.

In the multivariable logistic regression, which included 90,535 patients after excluding 3,590 whose prescriptions could not be uniquely assigned to a predefined opioid formulation group, other predictors also emerged. Increasing age was associated with lower odds of the absence of an opioid prescription, at a 1.4 percent decrease in odds per additional year. Male sex was associated with a 5.2 percent decrease in odds relative to female sex. Each additional point on the Charlson Comorbidity Index corresponded to a 3.4 percent decrease in odds, and having an assigned long-term care level (Pflegegrad) was associated with 30 percent lower odds. Strong opioid use showed one of the strongest inverse associations, with a 47.3 percent decrease in odds. Compared with exclusive use of long-acting opioids, exclusive use of short-acting opioids was associated with 36.7 percent lower odds, and combined long- and short-acting therapy with 19 percent lower odds. Psychotherapy was not significantly associated with the outcome.

The model showed moderate performance, with Nagelkerke’s R-squared of 0.204 and an area under the receiver operating characteristic curve of 0.722, indicating moderate discrimination. The authors caution that opioid discontinuation is a complex, highly patient-specific process that is difficult to capture in claims data, and that routinely available predictors explained only a proportion of the variation in the operationalized outcome. Multicollinearity between predictors was ruled out using a correlation matrix.

The finding on somatoform pain disorders carries particular guideline relevance. The German LONTS guideline on long-term opioid therapy for chronic non-cancer pain explicitly recommends against opioid use for persistent somatoform pain disorder (ICD-10 F45.40), and a German guideline on functional physical limitations states there is no evidence supporting opioid benefits in functional disorders, recommending withdrawal attempts within multimodal treatment plans where opioids have been initiated. Yet prior claims-data analyses cited by the authors indicate continued opioid exposure in this population, including one finding that 38 percent of patients with somatoform pain diagnoses received opioid prescriptions, with general practitioners as the primary specialists involved. The authors suggest their inverse association may reflect high symptom burden, psychosocial drivers of pain, and persistent prescribing despite limited expected benefits, rather than a favorable opioid response. Their data cannot determine whether opioid therapy was recently initiated or ongoing for years, nor whether opioid dependence influenced the results.

On adjunct therapies, the authors note that their result aligns with guideline frameworks positioning non-pharmacological care as central to chronic pain management. NICE recommends supervised exercise programmes for chronic primary pain, and a World Health Organization guideline for chronic primary low back pain highlights education, exercise, and physical therapies as core interventions. Evidence indicates exercise provides clinically meaningful benefits in chronic low back pain, though no single modality is clearly superior given trial heterogeneity, while evidence for passive modalities such as massage is more mixed. Consistent with prior German evidence that non-pharmacological interventions remain underused, particularly among older patients on long-term opioid therapy, nearly half the patients in this study received no adjunct therapy prescriptions alongside opioid treatment.

Several contextual limitations frame the conclusions. The data come from 2021, a period still affected by the COVID-19 pandemic in Germany, and pandemic-related changes in healthcare delivery may have influenced prescribing patterns. Non-opioid analgesics were not considered, multiple prescribers were not assessed, and the restriction to oral and transdermal formulations may have affected the representation of combination therapy and short-acting opioids. The authors also emphasize that the prescribing distribution should not be read as representative of opioid prescribing in Germany overall, since study-specific inclusion criteria shaped the population. They call for future studies with longer observation periods to confirm the associations, better verify discontinuation over time, and explore the underlying mechanisms, while highlighting the practical importance of psychosomatic assessment and systematic adjunct therapy use when managing opioid therapy for chronic pain in general practice.

Subject of Research: Predictors of opioid therapy discontinuation among primary care patients in Germany using 2021 claims data

Article Title: Opioid Therapy Discontinuation in General Practice: Associations with Adjunct Therapies and Somatoform Pain Disorders

Article References: Brinkmöller, S., Poß-Doering, R., Straßner, C., & Laux, G. (2026). Opioid Therapy Discontinuation in General Practice: Associations with Adjunct Therapies and Somatoform Pain Disorders. Drugs – Real World Outcomes, 13(3), 475-486. https://doi.org/10.1007/s40801-026-00573-7

Image Credits: AI Generated

DOI: 10.1007/s40801-026-00573-7

Keywords: opioids, chronic non-cancer pain, primary care, general practice, claims data, opioid discontinuation, adjunct therapies, somatoform pain disorders, substance use disorders, physiotherapy, Germany, observational study