OBJECTIVES: To assess the impact of pDPN/PHN-related health impairment on LOPT in patients treated with chronic pDPN/PHN. METHODS:  Using data from 777 employed adults with ≥3 months of pDPN/PHN and receiving pain medications, the effect of pDPN/PHN-related impairment on LOPT was estimated by: 1) single equation probit models (SEPM) assuming pain severity was exogenous, adjusting for respondents’ demographics, depression, anxiety, pain duration, type of pain, social isolation (felt alone/didn’t want to go out) and psychological distress (felt older/sleep difficulties); and 2) seemingly unrelated bivariate probit models (SUBPM), hypothesizing pain severity was endogenous and considering the same explanatory variables from the SEPM as instrumental variables (IVs). Pain severity was measured using a rating scale ranging from 0 (“no pain”) to 10 (“pain as bad as you can imagine”). RESULTS:  Thirty percent of respondents reported LOPT. Compared to respondents without LOPT, respondents with LOPT were younger, male, and had more moderate/severe pain (all p<0.001). Pain severity appears significantly related with social isolation and psychological distress. Its relationship with LOPT appears only significant when the latter variables were neglected in the SEPM or when they were used as IVs in the SUBPM. Marginal effects indicated that, compared to respondents with pain severity <4, those with pain severity ≥4 were 19.3% more likely to incur LOPT when using the SEPM and, 29.5% more likely when using the SUBPM (all p<0.001). Model specification tests suggested that pain severity to be truly endogenous and that all IVs jointly had sufficient explanatory power (all p<0.001). CONCLUSIONS: Pain severity has a significant impact on LOPT. The degree of this impact depends on ones’ ideas about how this impact is mediated by respondents’ social isolation and psychological distress. Alternative IVs for adjusting endogeneity bias of pain severity are worth exploring.