Cancer research output of North Africa did not contract after 2011 but turned inward: a controlled interrupted time-series and collaboration analysis, 1996 to 2025

Abstract

Background: Cancer research capacity across North Africa has grown unevenly, and the effect of Libya's post-2011 institutional disruption on its scientific output has not been measured. This study quantifies the Libyan trajectory against four regional comparators and tests whether it diverged after 2011.

Methods: All cancer-related records from Libya, Egypt, Tunisia, Algeria, and Morocco indexed in OpenAlex between 1996 and 2025 were retrieved (156,715 unique records). Annual counts were modeled with a controlled interrupted time series using negative binomial regression, with country fixed effects and cluster-robust standard errors, with Libya as the treated series and the interruption set at 2011. International co-authorship was modeled at the record level by logistic regression.

Results: Libyan output did not fall after 2011. The level change relative to the region was not significant (incidence rate ratio 0.88, 95% confidence interval 0.57 to 1.36), and the post-2011 slope exceeded the regional trend (incidence rate ratio 1.06 per year, 1.035 to 1.077). International collaboration diverged in the opposite direction: every comparator co-authorship rate rose after 2011, whereas Libya declined (odds ratio 0.73 per year, 0.65 to 0.82). The composition of Libyan collaboration also rotated: the share of international papers involving Western partners fell from 55 to 44 percent, while the regional Arab share rose from 25 to 59 percent, carried chiefly by a strengthening tie with Egypt. Contributing institutions per paper rose, indicating denser domestic networks rather than isolation.

Conclusions: Institutional disruption did not reduce the volume of Libyan cancer research, which continued to grow, but coincided with a turn toward domestic and regional rather than Western authorship. Quantity and integration moved in opposite directions, a pattern that volume-only bibliometrics would miss.