Assessing the Risk of Falls and Fractures in Older Adults with Diabetic Polyneuropathy Initiated on Gabapentinoids, SNRIs, or TCAs: An Observational Study (SAFER-PDPN)

Study type

• Study topic:
  Disease /health condition
• Study type:
  Non-interventional study
• Scope of the study:
  Safety study (incl. comparative)
• If ‘Not applicable’, further details on the study type:
  Obeservational study emulating a clinical trial.
• Data collection methods:
  Secondary use of data

Study drug and medical condition

• Study drug International non-proprietary name (INN) or common name:
  GABAPENTIN

  PREGABALIN

  AMITRIPTYLINE

  VENLAFAXINE

  DULOXETINE
• Anatomical Therapeutic Chemical (ATC) code:
  (N02BF01) gabapentin

  gabapentin

  (N02BF02) pregabalin

  pregabalin

  (N06AA09) amitriptyline

  amitriptyline

  (N06AA10) nortriptyline

  nortriptyline

  (N06AX16) venlafaxine

  venlafaxine

  (N06AX21) duloxetine

  duloxetine
• Medical condition to be studied:
  Fall

  Fracture
• Additional medical condition(s):
  Painful diabetic polyneuropathy (PDPN)

Population studied

• Short description of the study population:
  Persons aged 65 with diabetes initiating any of the studied drugs (proxy for painful diabetic polyneuropathy), with no treatment with any of the studied drugs in the last year and without diagnoses for depression, restless-legs syndrome, migraine, epilepsy, cardiac arrhythmia, heart block, congestive heart failure, urinary incontinence, falls or fractures in the last five years and no myocardial infarction in the last year.
• Age groups:
  • Adult and elderly population (≥18 years)

    • Elderly (≥ 65 years)

Study design details

• Study design:
  We will emulate a clinical trial of gabapentinoids versus SNRIs or TCAs in PDPN using the Danish nationwide registers. Randomisation will be mimicked using a set of pre-specified propensity score weights.
• Main study objective:
  1°: To estimate the risk for falls and fractures in patients with painful diabetic neuropathy after initiation of gabapentinoids versus SNRIs or TCAs.; 2°: To estimate the risk for falls and fractures in patients with painful diabetic polyneuropathy after initiation of gabapentin versus pregabalin, duloxetine, venlafaxine, amitriptyline or nortriptyline.; ;
• Setting:
  Danish national registries in the timeperiod from 01.01.2004 to 31.12.2024.
• Interventions:
  Initiation of gabapentinoids (pregabalin or gabapentin), selective noradrenalin reuptake inhibitors (duloxetin or venlafaxine) or tricyclic antidepressants (amitriptyline or nortriptyline).
• Comparators:
  Only active comparators.
• Outcomes:
  1°: Crude and propensity score weighting adjusted Hazard ratios (HRs) for falls and fall-related injuries for gabapentinoids versus SNRI or TCA initiators.; 2°: Crude and propensity score weighting adjusted Hazard ratios (HRs) for falls and fall-related injuries for initiators of gabapentin versus initiators of pregabalin, an SNRI or a TCA.
• Data analysis plan:
  First, we will determine, whether pregabalin and gabapentin are a suitable reference group (i.e. whether they are homogenous enough). We will do this by assessing the covariate overlap, crude and adjusted HRs. We will calculate propensity scores with logistic regression, where the exposure to either gabapentin or pregabalin will be the dependent variable and the predefined covariates the independent variables. We will use standardised mortality ratio weights (SMRWs) to estimate the average effect on the treated, with gabapentin as the reference. Should there be no clinically relevant difference, we will proceed with a drug class based comparison described below.; We will calculate propensity scores with logistic regression, where the exposure to any of the studied drugs will be the dependent variable (gabapentinoids are the reference and the respective other drug groups the comparator). Predefined covariates will be the independent variables. We will perform the analysis with propensity score weighting using SMRWs. For this, we will set the weights to 1 for gabapentinoid initiators and to calculated weights for TCA and SNRI initiators. As SMRWs can be susceptible to extreme weights, we will follow Stürmer, et al.’s approach to trim the weights at the 2.5th and the 97.5th percentile. Therefore, our estimand is the average effect on the treated (i.e., what would have been the fall and fracture risk had gabapentinoid initiators, instead initiated TCAs or SNRIs). We will analyse the effect of study drug initiation using a multivariate Cox regression model, estimating crude and propensity score weighted hazard ratios (HRs). We will use a sandwich (robust covariance matrix) estimator for the confidence intervals (CIs), to take into account the multiple inclusion of patients.; The analysis of for the secondary outcomes will follow the same procedure.