Structural drivers of drug shortages and market withdrawal: a longitudinal analysis of the Romanian pharmaceutical market

Abstract

Background

Drug supply discontinuities and market withdrawals represent a general structural weakness of national healthcare systems, with direct effects on therapeutic adherence and the equity of treatment access. This study investigates the longitudinal dynamics of shortages and drug withdrawal.

Methods

We constructed an annual dataset (2017–2024) from official national records and applied an Interrupted Time Series (ITS) analysis for shortages (as a rate adjusted to the number of Rx products), withdrawals, and SNAs, considering 2020 as the intervention point. To identify predictors of permanent shortages, we performed an observational analysis at the episode level (N = 603) and estimated a bias-reduced (Firth) logistic regression model. In parallel, we assessed the market longevity for 74,559 product-year observations (2008–2024), identifying 16,638 withdrawals and estimating a Cox model with time-dependent effects and frailty terms at the INN level.

Results

The ITS outcome points to a boost in shortages before the pandemic, a sharp drop in 2020, followed by a return to the upward trend with no statistically significant change in the post-COVID slope. The lag analysis suggests that episodes of shortage precede permanent withdrawals by an average interval of approximately two years. The Firth regression shows that generic status reduces the probability of permanent shortage compared to innovative products, and ATC M and N therapeutic classes are associated with a lower probability of permanence. In the survival analysis, the risk of market withdrawal increases with intensified competition within the same therapeutic group defined by INN and with higher prices for the same active substance, and generic and biosimilar products have a higher hazard of withdrawal than innovative products, with attenuation over time.

Conclusion

Drug shortages in Romania reflect market fragility that predated the pandemic. The results support the necessity of moving from reactive policies to predictive surveillance and price reforms.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-026-14893-5.

Background

Drug shortages pose a significant worldwide problem, leading to delayed treatments, increased system level costs, and adverse health consequences. These shortages interfere with routine clinical workflows, force clinicians to rely on unplanned alternatives, and exacerbate access disparities, particularly among vulnerable populations [1]. The consequences of drug shortages extend beyond care interruptions to include financial strain on patients and a spectrum of adverse outcomes, from medication errors and treatment modifications to fatal complications [2]. Research, predominantly from North America (70%), indicates that drug shortages typically result in a decline in medication usage, with observed reductions fluctuating between 1 and 99%. However, the diversity of data sources and analytical frameworks employed across these studies underscores a lack of standardized measurement for this phenomenon [3]. Shortages impact all patients but are especially detrimental to vulnerable groups like children. This pronounced effect stems from the specific medication dosages and heightened vulnerabilities inherent in pediatric care [4]. Between 2017 and 2019, shortage notifications jumped by 60% across 14 OECD members, which is a significant trend, even though differing national monitoring methods make such wide-scale analysis difficult [5]. The constant management of fluctuating drug availability strains frontline providers and threatens the consistency of care delivery. This suggests that pharmaceutical shortages are a symptom of a deeper structural vulnerability, directly impacting the healthcare system’s ability to remain efficient and equitable under pressure [6].

Current research of the drug shortage phenomenon frequently examines fragmented events or specific drug classes and often uses a descriptive approach [7–10], failing to provide a longitudinal perspective on the interplay between supply failures, market exits, and regulatory constraints. This is especially true in Central and Eastern Europe, where unique pricing frameworks and inadequate legal protections create hurdles not typically seen in Western healthcare systems [11]. This study addresses a critical gap in the health-services literature by providing the first longitudinal, system-level analysis of medicine shortages and market withdrawals in a Central and Eastern European context.

Supply disruptions for medicinal products are one of the most persistent and poorly understood weaknesses of modern healthcare systems [12], even with the rapid development of new technologies that are supposed to provide ways to manage crises that threaten both patients and public health systems [13]. Beyond their specific scope, these phenomena reflect structural imbalances between regulation [14], market maintenance mechanisms, and production capacity, with direct consequences for therapeutic continuity, equity in access to treatment, and the financial sustainability of public systems, which are under increasing pressure every year [15, 16]. Despite growing institutional interest in this issue at the European level [17, 18], most analyses remain descriptive and limited in time, providing a fragmented picture of the real dynamics and scale of discontinuities [19]. Discontinuities in drug supply are not an isolated phenomenon that occurs only at the national level, but a global public health problem [20] that impacts health systems in countries regardless of their level of economic development. Studies in the United States document the clinical and economic effects of treatment interruptions or substitutions due to shortages [21], while in China, the literature presents research on shortages of critical medicines, as defined by specific necessities leading to substandard medical care [22, 23]. India also presents supply problems from the perspective of inequality in access to effective medical care [24].

In Romania, the issue of drug shortages is frequently raised in public discourse and administrative reports, but systematic approaches based on longitudinal data and rigorous quantitative methods are rare [2]. Current analyses have primarily concentrated on discrete episodes or restricted product categories, neglecting the temporal aspect, isolating the structural factors influencing market withdrawals, and evaluating the correlation between supply instability and the progression of the life cycle of pharmaceutical products [19, 25, 26]. This has an impact on policymakers’ capacity to develop evidence-based policies and anticipate systemic risks [7]. In this context, the COVID-19 pandemic has been an unprecedented stress test for global supply chains while also providing a natural setting for assessing the resilience of the pharmaceutical market [27, 28]. However, interpreting the impact of the pandemic on discontinuities requires analytical tools capable of distinguishing between pre-existing trends and cyclical effects, avoiding the superficial attribution of structural system dysfunctions to a single external event [27, 29, 30].

Characterized by a strong import dependency, Romania’s pharmaceutical market features a diminishing yet significant local manufacturing sector, limited to generic versions of medicinal products [31]. Oversight is defined by a centralized and often restrictive regulatory framework. A key financial obstacle is the clawback tax, a quarterly contribution from MAHs designed to offset overspending relative to the national health budget [32]. Systemic bottlenecks continue to hinder the reimbursement pathway for new therapies, with relatively long delays and a substantial backlog of approved but non-reimbursed indications [33]. Compounding this landscape is the external reference pricing mechanism, which pegs local prices to lower levels in other EU member states, creating challenges for the supply chain reliability [34]. Drug shortages constitute an endemic challenge within the Romanian pharmaceutical sector, resulting in chronic supply chain instabilities that frequently impede access to essential medications across both community pharmacies and hospital settings [35]. Although the potential influence of regulatory policy on pharmaceutical availability is frequently debated, the magnitude and structural characteristics of these shortages remain underexplored in the current empirical literature.

In this study, we provide a comprehensive evaluation of medicine supply dynamics and shortages in Romania using a multi-faceted quantitative approach. First, we employed Interrupted Time Series (ITS) analysis to quantify temporal shifts in shortages, withdrawals, and SNA requests, specifically assessing how the COVID-19 pandemic altered existing levels and trends. Second, we conducted a retrospective, product-level observational analysis to identify factors distinguishing temporary from permanent shortages; this was supported by a bias-reduced (Firth) logistic regression model to address potential sparse data and separation issues. Finally, we analyzed the market longevity of 74,559 medicinal product-year entries from 2008 to 2024. By identifying 16,638 market withdrawals and applying a time-varying Cox proportional hazards model with a gamma-distributed frailty term, we evaluated how competition, regulatory status, and pricing influenced the likelihood of market exit.

Materials and methods

Data sets

To analyze medicine supply dynamics and shortages in Romania between 2017 and 2024, we constructed a time-series dataset utilizing official, publicly available national indicators on medicine supply in Romania [36]. The data for each calendar year provides national aggregate figures for the following four variables: a) frequency of reported medicine shortages to the national authority; b) volume of permanent market discontinuations (withdrawals); c) an estimation of the overall market size based on the count of marketed prescription-only medicines; d) the number of Special Needs Authorizations (SNA) granted for emergency access to otherwise unavailable medicines [37]. To account for differences in measurement scale, shortages were analyzed both as raw counts and as rates per marketed prescription product (the rate used the total number of Rx medicines as a denominator). The final analytic dataset consisted of eight consecutive annual data points (2017–2024) for all indicators.

We analyzed data from 12 editions of CaNaMed, the official national price catalogue for medicines in Romania, in order to conduct a drug survival analysis and determine which factors are associated with the withdrawal of medicinal products from the market. The editions included were: 2008–2015, 2018, 2020, 2023, 2024, and 2025. Each edition of the catalog was retrieved from the official Ministry of Health website [38] during its respective period of validity. Individual catalog files were imported into R [39] and processed through a standardized cleaning and harmonization pipeline to ensure consistent programmatic extraction of key variables.

The original product status field in the national price catalog contained numerous heterogeneous labels used across time by the legislation and the health authorities’ practice. For analytical consistency, these raw labels were harmonized into four mutually exclusive categories: a) innovative (included all originator products, identified through labels such as “innovative”, “pioneer”, and “orphan”); generic (included all multisource small-molecule products, mapped from labels such as generic, or “first generic”); biologicals_multisource (included biosimilars and other non-originator biological products, mapped from labels such as biosimilar and blood derived); others (included product types that did not fit the above categories, such as vaccines, homeopathic products, and herbal products). The whole data set consisted of 74,559 observations. Complete-case analysis was applied to the dataset; observations with a wholesale price of zero (n = 575) were treated as non-recoverable and excluded, with no imputation performed.

Interrupted time-series design

Using an Interrupted Time Series (ITS) approach, we quantified temporal changes in shortages, withdrawals, and SNA requests, assessing if the onset of the COVID-19 pandemic led to alterations in level or trend. The intervention point was set at 2020, reflecting major disruptions across regulation, logistics, and clinical operations during the first wave.

A segmented regression model was fitted for each outcome, explicitly allowing for distinct slopes before and after COVID-19 (i.e., year 2020), with the form:

where time is a continuous variable (1–8), COVID is an indicator taking the value 1 from 2020 onward, and time_after_COVID represents the number of years elapsed since the 2020 intervention and allows the model to estimate whether the rate of increase or decrease in the outcome changed in the post-COVID period relative to the pre-COVID trend. Using this model, it becomes possible to distinguish between the baseline pre-pandemic trend, the immediate impact of COVID-19‘s onset, and any sustained change in the trend’s direction or steepness afterward.

Since shortages, product withdrawals, and SNAs are recorded as counts, we used Poisson regression, which is a standard method for such data. For shortages specifically, we adjusted for the total number of medicines available on the market (using an offset term equal to the log of the total number of prescription medicines on the market), allowing results to be interpreted as shortage rates. To check for overdispersion, we analyzed the ratio of deviance to degrees of freedom. Since overdispersion was detected, we re-estimated the models using Negative Binomial regression for validation. For each model, results are expressed as regression coefficients and Incidence Rate Ratios (IRRs), accompanied by 95% confidence intervals. The IRRs are obtained by exponentiating the estimated coefficients. We explored whether structural market exits (withdrawals) predicted subsequent shortages by estimating a lagged Poisson model. This approach modeled the count of shortages in a given year (t) based on the total number of withdrawals from the preceding year (t-1). We adjusted for potential confounders by controlling for both the overall time trend and the total number of marketed products. Additionally, we utilized cross-correlation analysis to provide a non-parametric view of the relationship’s timing and strength. The limited time span prompted a cautious approach to uncertainty: HC0 sandwich estimators were used for the Poisson regressions as a sensitivity check, and Newey–West standard errors were used with the log-rate specifications to control for potential autocorrelation. Deviance scores, Akaike Information Criterion (AIC), and visual comparisons of projected and counterfactual trends were used to assess model fit quality. To facilitate the interpretation of the ITS results, we generated a line plot displaying observed yearly data, the model-fitted trend, and a counterfactual line in the absence of COVID-19.

Determinants of shortage permanence

We performed a retrospective, product-level observational analysis aimed at identifying factors linked to whether medicine shortages were permanent or only temporary. The unit of analysis was the shortage episode, classified as temporary or permanent based on regulatory documentation issued by the National Agency for Medicines and Medical Devices from Romania (NAMMDR). Each observation consisted of a unique combination of brand name, active substance, strength, pharmaceutical form, MAH, and calendar year. The reference category was established using temporary shortages, given the high probability that these events stem from idiosyncratic, short-term disruptions (e.g., raw material constraints or logistical impediments). Conversely, the persistence of permanent shortages suggests an underlying erosion of market stability, frequently associated with long-term structural issues, including the loss of economic viability or manufacturer decisions regarding commercial and regulatory adherence.

We pre-specified the independent variables based on regulatory relevance, data completeness, and interpretability. The following were included in the model:

• Wholesale price (expressed as a continuous variable in RON);

• Route of administration, treated as a categorical variable;

• Therapeutic class, defined using the first level of the Anatomical Therapeutic Chemical (ATC) classification;

• Innovation status, defined as generic, innovative, or orphan medicine (according to Romanian pricing legislation, specifically Order of the Minister of Health No. 368/2017) [40];

• Calendar year of shortage, included as a continuous variable to analyze temporal trends.

We did not include over-the-counter (OTC) observations because, in line with the Romanian legal framework, there are no officially public or uniform pricing data for OTC, retail margins vary from pharmacy to pharmacy, and wholesale costs are also freely established and cannot be inferred in a reasonable way. The sterile or non-sterile characteristic of medicines was not included in the model, as it has a structural collinearity with the route of administration, and the marketing authorization procedure (national or mutual recognition procedure/decentralized procedure, MRP/DCP) was excluded because it is not systematically reported in the official documents.

We developed multivariable logistic regression models to investigate the relationship between certain product characteristics and the likelihood of a shortage being permanent. Preliminary analyses with standard and quasibinomial logistic regression uncovered issues of sparse data categories and quasi-separation, particularly in less common routes of administration and less represented ATC classes. To address these issues, the primary analysis utilized bias-reduced logistic regression with Firth-type penalization, implemented through adjusted score equations.

This technique tends to minimize bias in small samples (we used the median bias-reducing score adjustments in Kenne Pagui et al. (2017) [41] as implemented in the brglm2 R package [42]) and ensures reliable estimates even under separation conditions. All models used a logit link and were estimated via maximum penalized likelihood. Regression results were expressed in terms of odds ratios (ORs) with 95% confidence intervals, and statistical significance was determined using two-sided tests at a significance threshold of 0.05.

To assess model stability, we examined both coefficient magnitudes and standard errors. We also tested for separation before applying the penalized estimation. For model discrimination, we used the AUC (area under the receiver operating characteristic curve), which was calculated on the estimation sample after excluding cases with missing covariate data. Deviance statistics and the Akaike Information Criterion (AIC) were also used to evaluate model fit.

We conducted sensitivity analyses by comparing the bias-reduced model with a quasibinomial logistic regression. Robustness was evaluated based on whether coefficients maintained consistent directionality and levels of statistical significance between methods.

Drug survival analysis

The survival time for each medicinal product on the Romanian pharmaceutical market was defined as the number of years from its reference date (the earliest year in which the product was observed in the catalog) until either market withdrawal, operationalized as permanent discontinuation of reimbursement or listing, or censoring, in cases where the product remained listed at the end of follow-up in 2025.

Survival analysis was performed using the Cox proportional hazards regression model to assess the association between selected covariates and the time to market exit, with right-censoring for observations that did not experience the event by the end of follow-up. All analyses were conducted in R (version 4.5.2) using the “survival” package. An initial base model was fitted including the following fixed covariates: the natural logarithm of price, competition within the same international non-proprietary name (INN, expressed as the number of competing companies per INN in the relevant year), a four-category regulatory status variable (innovative, generic, biologicals multisource, and others), and a pharmaceutical form category variable (oral, cutaneous/topical, parenteral, others). The proportional hazards (PH) assumption was evaluated using Schoenfeld residuals with a Kaplan-Meier transformation, revealing significant violations for the status and pharmaceutical form (p < 0.006). To address PH violations, time-dependent effects were introduced for these covariates using log-time interactions. The final model incorporated two time-invariant terms (competition, and log(price)), and time-dependent terms (interactions between regulatory status and pharmaceutical form category with log(time)). The model was fitted using the coxph function in R, with goodness-of-fit assessed via likelihood ratio tests and residual diagnostics.

All p-values are two-sided, and statistical significance was considered at α = 0.05 unless otherwise stated.

Software

All analyses were performed in the R computing and programming environment (version 4.5.2) [39] using the dplyr [43], ggplot2 [44], broom [45], MASS [46], lmtest [47], sandwich [48], AER [49], modelsummary [50], and “survival” [51] packages. Bias-reduced logistic regression was implemented using the brglm2 package [42], and model diagnostics were conducted using the pROC [52], car [53], and stats [39] packages.

Results

Interrupted time-series design

Using interrupted time series analysis, we analyzed yearly drug shortage data from 2017 to 2024 to determine how the COVID-19 pandemic affected shortage patterns, with 2020 designated as the intervention point. The primary Poisson model, which controlled for the total number of products available in the market, revealed that drug shortages were already rising significantly before the pandemic began (time coefficient β = 0.292, p < 0.01). Contrary to expectations, COVID-19‘s immediate effect was a rather significant decrease in shortages (β = −0.889, p < 0.001). Following the pandemic’s onset, shortages resumed an upward trajectory, though this post-pandemic trend shift was not statistically significant (β = 0.138, p = 0.25) (Table 1).

Table 1.

Interrupted time series models of drug shortages, withdrawals, and ANS (2017–2024). Coefficients show pre-pandemic trends, COVID-19 level changes, and post-COVID slope changes

	Poisson ITS	NegBin ITS	Withdrawals ITS	ANS ITS
(Intercept)	−5.560***	−5.532***	6.334***	4.268***
	(0.249)	(0.370)	(0.059)	(0.183)
Time	0.292**	0.276+	0.130***	−0.014
	(0.108)	(0.167)	(0.026)	(0.085)
Covid	−0.889***	−0.735*	0.713***	0.243
	(0.187)	(0.314)	(0.045)	(0.151)
Time after COVID	0.138	0.116	−0.366***	0.102
	(0.114)	(0.180)	(0.028)	(0.090)
Num.Obs.	8	8	8	8

+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001

Poisson dispersion tests revealed strong overdispersion (an alpha value of 2.2 using the dispersiontest () function from the AER library), supporting the use of negative binomial sensitivity analyses. A negative binomial regression, applied to mitigate the effects of overdispersion, had results consistent with those of the first model. The pre-pandemic trend continued to show a positive effect (β = 0.276, p < 0.1), while the immediate impact of COVID-19 remained significantly negative (β = −0.735, p < 0.05). The post-COVID slope change, however, was statistically nonsignificant (β = 0.116, p = 0.53), suggesting that overdispersion had minimal influence on the overall interpretation.

Figure 1 displays the observed shortage counts, the fitted ITS model trajectory, and the counterfactual scenario (projected shortages had the pre-pandemic trend continued without a COVID-19 effect). In 2020, the recorded shortages fell below the counterfactual trajectory, a pattern consistent with the statistically significant negative level shift estimated at the pandemic onset. From 2021 onward, the observed shortages rebounded and followed a trend similar to the counterfactual, reflecting the non-significant slope change in shortage incidence post-COVID.

Fig. 1.

Observed drug shortages (gray bars) with fitted Poisson ITS estimates (solid green line and points) and counterfactual predictions assuming no COVID-19 intervention (dashed orange line). The vertical dashed line corresponds to the COVID-19 intervention year (2020)

Poisson ITS models were separately fitted for drug withdrawals and ANSs. Withdrawals increased steadily over time (β = 0.130, p < 0.001) with a sharp rise in 2020 (β = 0.713, p < 0.001). Yet, the slope after COVID decreased significantly (β = −0.366, p < 0.001), suggesting a slowdown in withdrawal rates post-pandemic onset. For ANS, no statistically significant changes were detected immediately after COVID-19 or in the trajectory, as trends remained essentially flat (Table 1).

The robustness investigation was not limited to swapping Poisson for negative binomial; we used two additional tests. Lag analysis indicated that the strongest correlation was at lag −2. In other words, a high number of reported shortages seems to be followed by a high number of withdrawals after about two years. This gap suggests an effect of the regulatory framework or of the built-in decision cycle, where a period of supply instability (shortage) results after two years in an administrative decision. Robust small-sample standard errors (sandwich/HC0 and Newey–West corrections) yielded significance patterns consistent with the standard Poisson estimates, underscoring the robustness of the findings. When employing Newey–West robust standard errors, which account for autocorrelation and overdispersion, the pre-COVID trend lost statistical significance, which is not surprising, considering the more conservative character of this statistical approach. However, the immediate effect of the COVID-19 onset remained statistically significant. This confirms the robustness of the finding that the pandemic caused a genuine, abrupt, short-term change in medicine availability.

Determinants of shortage permanence

We also investigated the factors associated with drug shortage types (temporary versus permanent) using a bias-reduced (Firth) logistic regression model, which accounted for potential sparse data and separation issues. The Firth logistic regression model converged after 9 iterations. Residual diagnostics did not indicate extreme deviations (only two observations had residuals below −2, which were inspected and found to be valid extreme cases, so they were retained in the analysis), and variance inflation factors suggested no concerning multicollinearity among predictors. Sensitivity analyses, including alternative coding of categorical predictors and comparison with a standard logistic model, yielded similar odds ratios, indicating that the results are robust. The results of the bias-reduced (Firth) logistic regression are summarized in the forest plot (Fig. 2), which displays the odds ratios and 95% confidence intervals for each predictor. Our analysis suggests that therapeutic class and innovation status were the strongest predictors of shortage type. Specifically, drugs in the musculoskeletal (ATC class M) and nervous system (ATC class N) categories showed significantly lower odds of being subject to a permanent shortage. Other ATC classes, including C (cardiovascular products) and H (systemic hormonal preparations, excluding sex hormones and insulins), displayed borderline associations, while the remaining classes did not show statistically significant effects. These disparities likely reflect sector-specific differences in production stability and market structural integrity.

Fig. 2.

Forest plot of odds ratios from the bias-reduced (Firth) logistic regression model predicting shortage type (N = 603). Odds ratios (ORs) and 95% confidence intervals are shown for each predictor. The red dashed line indicates OR = 1, representing no association. ORs < 1 indicate lower odds of shortage relative to the reference category, and ORs > 1 indicate higher odds

Innovation status emerged as a significant predictor, with generic drugs exhibiting substantially lower odds of permanent shortages compared to innovative products (OR 0.53, 95% CI 0.33–0.85), which means that when controlling for other covariates, generic medicines have about half the odds of experiencing a permanent shortage relative to their reference products. In contrast, no statistically significant differences were observed for biosimilars or orphan drugs when compared to the innovative reference group. Neither route of administration (p-values ranging from 0.12 to 0.65 across different routes) nor price (p = 0.29) showed statistically significant associations with the outcome after covariate adjustment, indicating limited predictive value for shortage type. A modest temporal trend suggested higher odds in more recent years, though this finding did not reach statistical significance (p = 0.06). The bias-reduced model demonstrated acceptable discrimination, with an area under the receiver operating characteristic curve (AUC = 0.71), indicating moderate ability to distinguish between permanent and temporary shortages. Model fit statistics showed a substantial reduction in deviance relative to the null model (AIC = 773.7). Results from standard and quasibinomial logistic regression models were directionally consistent with the primary bias-reduced analysis but exhibited inflated standard errors and evidence of quasi-separation, particularly for infrequent routes of administration and certain ATC classes. In these models, several coefficients displayed extreme values with large uncertainty, limiting interpretability. The key findings (the reduced odds of permanent shortages for ATC groups M and N, the protective effect of generic status, and the positive temporal trend) remained stable across modeling approaches. The bias-reduced model yielded more conservative and numerically stable estimates, supporting its use as the primary analytical framework.

Taken together, the results indicate that shortage permanence is strongly associated with therapeutic class and innovation status, while price and administration route do not seem to significantly influence the permanent shortage risk of a product. The observed temporal trend suggests increasing structural vulnerability in the pharmaceutical market over time. Sensitivity analyses confirmed the robustness of these findings.

Drug survival analysis

We examined the market longevity of 74,559 medicinal product-year entries in Romania from 2008 to 2024, identifying 16,638 instances of market withdrawal. To evaluate how competition, regulatory status, pharmaceutical form, and pricing influence the likelihood of exit, we applied a time-varying Cox proportional hazards model with a gamma-distributed frailty term for INNs.

Within the same INN, higher pricing was significantly associated with a greater risk of market exit. Each one-unit increase in the log-transformed price corresponded to a 17.5% higher hazard (HR = 1.175; 95% CI: 1.158–1.192; p < 0.001). More intuitively, this HR value suggests that doubling a medicinal product’s price relative to others with the same active ingredient increases the hazard of exit by approximately 12.6%.

Within the same INN cluster, the hazard of market exit was also significantly associated with the market competition (expressed as the number of companies having in the same year a product within the same INN); each additional competing company per year increased the hazard by 7.2% (HR = 1.072; 95% CI: 1.066–1.078; p < 0.001). Compared with innovative reference products, multisource biologicals had a nearly two-fold higher hazard (HR = 1.998, 95% CI: 1.589–2.511, p < 0.001), generic products had a 2.5-fold higher hazard (HR = 2.500, 95% CI: 2.299–2.718, p < 0.001), whereas other products showed the highest hazard (HR = 4.333, 95% CI: 3.748–5.009, p < 0.001). Time-varying estimates revealed a significant attenuation in exit risk over the study period (Fig. 3). Multisource biologicals’ hazard ratio decreases from 1.998 after 1 year to about 1.80 at 2 years and 1.70 at 3 years. For generics, the decline is steeper, from 2.500 at 1 year to roughly 1.98 at 2 years and 1.72 at 3 years. Hazard ratios for products with other regulatory statuses remain fairly unchanged over time. Within the same INN, the four dosage form categories showed minimal and statistically nonsignificant baseline effects on hazard. Additionally, time-varying analyses revealed no meaningful changes in hazard over time for pharmaceutical forms.

Fig. 3.

Time-varying hazard ratios for market exit by regulatory product status. Hazard ratios over time are displayed for multisource biological, generic, and other products relative to innovative products (reference HR = 1, dashed line). Shaded areas correspond to 95% confidence intervals. Estimates are conditional on INN (DCI) frailty. Negative time-varying effects for multisource biological and generic products indicate attenuation of risk over time, whereas the effect for other products remains stable. Time is shown in years of observation in the study period

The estimated variance of the gamma-distributed frailty term for INN was 0.985, indicating substantial unobserved heterogeneity in market exit hazards across INNs. In other words, products within the same INN share correlated exit risks that are not fully explained by our covariates. This is to be expected, because INNs corresponding to different therapeutic areas and applications vary widely in their perceived importance by the public, medical profession, health authorities, and even companies themselves. By incorporating this frailty term, the model effectively controls for INN-level variance, thereby refining the precision of our primary predictors and providing a more robust assessment of risk within the different INN groups.

The model showed strong discriminative ability, with a concordance index of 0.74 (SE = 0.003). A highly significant likelihood ratio test (χ² = 12,702, df = 1,803, p < 0.001) confirmed that the included covariates meaningfully explained variation in market exit events.

Discussion

The results of the interrupted time series analysis indicate that drug shortages in Romania have been on a marked upward trajectory since the pre-pandemic period. This finding points to the existence of previously existing structural vulnerabilities within the pharmaceutical market, which cannot be attributed entirely to the shock caused by the COVID-19 pandemic [54, 55]. The substantial decrease in the number of discontinuities reported in 2020 contradicts the assumption that the pandemic would automatically lead to an immediate worsening of the drug shortage. One possible explanation may relate to the administrative and regulatory interventions swiftly adopted in the context of the COVID-19 crisis. These included temporary export restrictions, prioritization of essential medicines supply, and procedural flexibilities. Such measures were implemented on a large scale across Europe, including in Romania, as part of rapid response strategies to avoid healthcare disruptions [56]. The literature shows that in many countries immediate control measures were adopted, such as trade restrictions on intra-EU deliveries, facilitation of imports, or adjustments to pharmaceutical practices [27]. The aim of such measures was to stabilize the availability of medicines in the short term [57]. These interventions were associated, in some cases, with a temporary stabilization of the availability of certain medicinal products. At the same time, some empirical analyses suggest that, following an initial peak in supply disruptions during the first months of the pandemic, their level subsequently declined. This suggests a rather transient shock followed by a relatively rapid adaptation of supply systems [29, 58, 59]. Taking all these into account, this interpretation should be viewed with caution, in the absence of direct evidence for the observed effect. Alternative explanations cannot be ruled out, for instance delays in reporting discontinuities due to administrative overload, temporary shifts in demand for medicines across different therapeutic classes, or artifacts created by changes in reporting practices. In this regard, the decline observed in 2020 could reflect not necessarily a real improvement in availability but also cyclical effects or limitations of the available data, consistent with the literature describing the pandemic as a stress test that amplified pre-existing weaknesses in pharmaceutical supply chains.

The lack of a significant change in the post-pandemic rate indicates the persistence of economic and regulatory factors that continue to negatively influence the sustainability of the Romanian pharmaceutical market. In this sense, the results support the perspective that global crises act more as amplifiers of pre-existing dysfunctions than as primary causes [60].

The finding that repeated episodes of shortage are followed, after an average interval of approximately two years, by the definitive withdrawal of products from the Romanian market indicates the existence of a sequential and cumulative process of decline in commercial viability. This relationship should be interpreted primarily as a temporal association rather than a direct causal link, given the observational nature of our data. This temporal correlation suggests that discontinuities are not simply temporary disruptions, but early signs of structural instability which, in the absence of effective intervention, can lead to the permanent withdrawal of products. However, it is possible that both shortages and withdrawals reflect common underlying drivers, such as economic pressures, regulatory constraints, or strategic decisions by MAHs. This interpretation is supported by the literature, which highlights that repeated shortages reflect the progressive cumulative effect of economic, operational, and regulatory pressures on manufacturers, leading to a reassessment of the commercial viability of products, particularly in the context of the high costs associated with restarting production and regulatory compliance [61, 62]. Published studies also indicate that manufacturers tend to withdraw products characterized by operational instability and low profitability, even when they are therapeutically relevant, particularly in small and highly regulated markets where commercial margins are limited by legislation and subject to over-taxation, as well as vulnerable to fluctuations in demand or supply chain disruptions [63, 64]. The time interval identified between the onset of discontinuities and the actual withdrawal is consistent with international literature descriptions of the gradual decision-making processes of pharmaceutical companies, which involve iterative assessments of costs, risks, and market prospects before initiating definitive withdrawal [12, 65]. In addition, the role of the national institutional framework is relevant in this process, as complex administrative procedures and the lack of automatic mechanisms for removing non-functional products can prolong the period between the emergence of discontinuities and formal withdrawal, leading to an underestimation of the real risk of unavailability. This situation, also documented in comparative European analyses, favors the formal maintenance on the market of products with virtually no access for patients, delaying the recognition of their ineffectiveness [66, 67] Overall, the results support the conclusions of the literature that repeated discontinuities are not accidental events, but manifestations of a complex process of erosion of economic sustainability, generated by the interaction between regulatory constraints, competitive dynamics, and structural limitations of the pharmaceutical market [14, 19].

Bias-reduced logistic regression analysis indicates that innovative status and therapeutic class are the main predictors of the permanence of discontinuities. Generic drugs have a significantly lower risk of being permanently withdrawn compared to innovative products. This result can be explained by the existence of several alternative manufacturers for generics, which increases supply redundancy and reduces dependence on a single supplier. Previous studies have shown that multi-source markets are more resilient to temporary disruptions [68, 69]. In contrast, innovative products are more vulnerable to withdrawal in the context of price pressures and low sales volumes, especially in strictly regulated markets such as Romania. These findings are consistent with the literature on the impact of external reference pricing and price capping policies on the availability of patented products [68, 70]. The lack of a significant association between price and the permanence of discontinuity suggests that structural and institutional factors may prevail over purely economic mechanisms in determining withdrawals.

Simultaneously, the results also indicate that therapeutic class assignment is an important determinant of discontinuity patterns, suggesting significant structural differences between segments of the pharmaceutical market. In particular, drugs in the musculoskeletal (ATC M) and nervous system (ATC N) classes were significantly less likely to be subject to permanent discontinuities than other therapeutic categories. This relative stability can be explained by the constant and predictable demand associated with common chronic conditions, such as rheumatic diseases, chronic pain, neurological and psychiatric disorders, which generate high consumption volumes and sustained commercial interest for manufacturers. The literature suggests that segments characterized by long-term use and large patient bases tend to benefit from greater supplier redundancy and continued investment in production capacity, contributing to increased supply resilience [23, 71]. In addition, these classes include many mature generic products with well-established supply chains, which reduces vulnerability to temporary disruptions [72]. In the opposite direction, the limited associations seen for cardiovascular (ATC C) and systemic hormonal (ATC H) classes might show more exposure to production, regulatory, and cost constraints, especially for substances with complex manufacturing processes or fluctuating demand. Previous studies indicate that cardiovascular and hormonal drugs are often affected by intense price pressures and international reference pricing policies, which may reduce commercial attractiveness in relatively low-volume markets, favoring the gradual withdrawal of some products [26, 73]. The absence of significant associations for other therapeutic classes suggests that, in these segments, discontinuities are driven more by circumstantial and product-specific factors than by systematic structural vulnerabilities. Overall, these results highlight the heterogeneous nature of the risk of discontinuity between different therapeutic areas and underscore the need for differentiated policies tailored to the particularities of each segment, rather than uniform interventions applied to the entire pharmaceutical market [74]. Future studies could go further and incorporate usage data at the ATC class level to better understand the relationship between how much a drug is used and how often supply disruptions occur. This approach would make possible to make an important distinction: are disruptions frequent simply because the volume of certain product use is high, or are there therapeutic classes where they shortages simply reflect deeper vulnerabilities in the supply chain? Furthermore, monitoring the evolution of these indicators over a longer period of time could contribute both to validating the relationships identified in this study.

The observation that each additional competitor raises the risk of market exit by 7.2% reveals a striking contradiction in how pharmaceutical markets function. While generics are generally assumed to enhance affordability and availability, the data indicate that in certain INN groups, too much competition can undermine market stability and drive companies out. This aligns with well-documented evidence that generic markets frequently experience “race-to-the-bottom” pricing pressures, in which accelerated competition steadily erodes profitability to the point of instability [75, 76]. Insufficient margins can force firms, particularly those lacking substantial economic buffers, to withdraw from the market, which in turn heightens market concentration. This is rather concerning, because when the supplier landscape becomes narrower and less diversified, it becomes more vulnerable to manufacturing problems, ultimately amplifying the likelihood of (possibly long-lasting) drug shortages. This is not necessarily related to a complete lack of profitability, but rather to low profitability and competing priorities for the same manufacturer [77].

These results can also be interpreted in the specific context of public procurement mechanisms in Romania, particularly national tenders and procurement procedures organized by hospitals, where the minimum price criterion plays a dominant role. The frequent practice of awarding contracts at very low price levels leads to the supply of products over long periods at prices that are often below the economic profitability threshold. Under these conditions, the cumulative lack of profitability, combined with long-term contractual liabilities, may lead to a reassessment of the commercial strategy by manufacturers and distributors, favoring either a gradual withdrawal from certain market segments or a refusal to participate in procurement procedures considered unsustainable. Thus, competitive pressures and procurement policies predominantly oriented towards cost minimisation may indirectly contribute to supply instability and increased risk of discontinuity, supporting the interpretation that price formation and contracting mechanisms are relevant structural factors in the dynamics of market withdrawals [78, 79]. At the same time, interpreting the results of this analysis requires taking into account the characteristics of each therapeutic class and products being evaluated. The effects of competition on market differ significantly from one case to another, depending on the therapeutic importance of the drug, the availability of appropriate therapeutic alternatives, the product origin, and the potential for substitution with an equivalent alternative [80]. In certain situations, increased competition brings clear and easily perceived benefits, for instance patients gaining easier access to the treatments they need, at significantly lower prices [81]. However, in segments where price pressure is already very high, especially for generic and biosimilar medicines, pricing mechanisms can trigger a gradual and difficult-to-stop erosion of prices [82]. In Romania, regulations require that the prices of these products be set below the level of reference medicines [40], which, in a context of intense competition and low sales volumes, can squeeze profit margins to levels where the business is no longer economically viable.

An important point to note is that these results reflect dynamics at the level of individual products (i.e., marketing authorizations), where the effects of competition may differ from what we observe when analyzing an entire therapeutic class. In other words, greater competition reduces the risk of patients being left without treatment options at the class level, simply because alternatives exist, yet the situation becomes more complicated when we examine each product individually. Some companies adjust their supply strategies to cope with the pressure, while others choose to withdraw completely from the market, all depending on whether profit margins still allow them to continue to keep the product on the market.

Thus, what initially appears to be an advantage: lower prices and easier access, can, over time, become a source of problems for drug availability. It is, in a way, the “flip side of the coin” of policies focused exclusively on price reduction. This is exactly why the findings of this study call for a more nuanced approach to pharmaceutical policies, one that treats market segments with different dynamics separately and strikes a real equilibrium between the goals of accessibility and long-term sustainability.

The association between relatively high prices and a higher probability of market withdrawal may seem counterintuitive, but it likely reflects the difficulty of maintaining competitiveness in the context of price control policies and international benchmarking [82, 83]. As is the situation in Romania, a relatively high price implies a cap on commercial margins enforced by the pricing regulations in force [40].

The significantly higher risk of exit for generic and multi-source biologic products, particularly in the first years after market launch, is consistent with studies describing shorter commercial life cycles for these categories, characterized by rapid market entry and accelerated exit [69].

The pronounced variability at the INN level highlights the influence of specific therapeutic and clinical factors on commercial stability, underscoring the need for differentiated approaches in regulatory policies. The results suggest that reactive policies focused exclusively on managing specific crises are insufficient to ensure supply stability. The absence of systematic mechanisms for assessing the risk of withdrawal limits the ability of authorities to intervene preventively. Predictive monitoring tools need to be developed, integrating economic, structural, and temporal indicators for the early identification of vulnerable products. Similar models have been proposed in other jurisdictions as part of pharmaceutical security strategies [20]. The results are also consistent with the argument that pricing and reimbursement policies should be reconsidered to maintain a long-term balance between affordability and economic sustainability.

To better understand these findings, it’s worth considering the broader national context. Several public policy measures are already in place to manage and limit drug shortages, though they tend to focus on monitoring and responding to potential emergencies, rather than offering a comprehensive, tiered system built around prevention. As part of the current Romanian law, pharmacies, wholesalers, and MAHs must use the Electronic Reporting System (ERS) to report their stocks and operations every day [84]. Beyond reporting, MAHs and distributors are also obligated, as part of their public service duties, to ensure that the medicines in their responsibility are consistently and adequately supplied [85]. In terms of trade flows, any shipments within the EU must be flagged at least 10 business days in advance. Parallel trade is only permitted when there are enough stocks on hand to meet national demand, which is calculated based on average turnover figures [85]. There have been multiple orders of the minister of health along time that temporarily suspended the parallel export for a list of medicinal products. Also, the rules in force include obligations to notify discontinuities, in accordance with Law No. 95/2006, which requires shortages to be announced at least 6 months in advance (or 12 months in the case of shortages due to commercial reasons) [86]. These rules are based on the provisions of Directive 2001/83/EC [87], which establish a minimum period of 2 months, save in exceptional circumstances. When a product is completely unavailable on the market and cannot be obtained through the usual channels, Romanian authorities may use alternative access mechanisms, such as special needs authorizations [88] or marketing authorizations for medicines on public health grounds [89], which benefit from accelerated approval procedures and are intended exclusively for the Romanian market, to ensure the continuity of treatments. Although the pricing system is built around an external benchmarking mechanism based on the lowest price in a basket of 12 EU countries [40], the legislative framework allows some flexibility in specific contexts. These include cost-volume and cost-volume-outcome contracts, the use of alternative pricing methods in the absence of international comparators, as well as differentiated treatments for certain categories of medicines, such as orphan or pediatric products [40]. However, such mechanisms are applied on an ad hoc basis and do not constitute an integrated prevention system, as they are activated primarily in response to specific issues of access or availability. The results of our study suggest that these tools, while necessary, are predominantly responsive rather than precautionary. At the same time, they do not directly address the factors causing market insecurity, such as price pressuring, competitive behavior, or manufacturers’ business planning.

Romania’s findings have implications far beyond its borders, as its pharmaceutical market mirrors the structural characteristics of several EU member states that employ rigid external reference pricing and intensive cost-containment measures. The Romanian model utilizes one of the most stringent price benchmarks in the European Union, setting prices based on the absolute minimum across a 12-country basket [11, 90]. This specific regulatory mechanism is strongly linked to a higher probability of market exits for low-margin generic drugs and creates a systemic vulnerability to parallel exports [11, 34]. Consequently, the Romanian experience serves as a critical case study for understanding how aggressive pricing policies can inadvertently compromise the long-term availability of essential medicines across similar European health systems. Comparable supply-side pressures have been recorded across Europe, characterized by a manyfold surge in shortage incidents over the previous two decades. Data indicates that these disruptions are no longer isolated events but are systematically reported across all surveyed EU nations. This dramatic longitudinal increase underscores a growing instability in the European pharmaceutical market, shifting the problem from a localized issue to a persistent, continent-wide systemic failure [26, 91]. Determinants such as rigid price controls, insufficient domestic production, and the influence of international trade are common features not only in Romania but across other small, cost-sensitive markets in Central and Eastern Europe [92]. Furthermore, the European Court of Auditors has observed that shortages across the Union are the product of systemic flaws, including disjointed reporting, inadequate early-detection systems, and regulatory inconsistencies, rather than being limited to the failures of individual nations [93]. Therefore, the relationship between long-term supply instability and subsequent market exits identified in this study is likely representative of broader trends within the EU, especially in countries with similar pricing and regulatory constraints. This suggests that the “shortage-to-withdrawal” pipeline is a structural risk shared by multiple European health systems, making the evidence from this study a valuable proxy for understanding supply chain fragility across the continent.

While the outcomes of this study can be understood in relation to the broader regulatory framework at the European level, the specific ways in which shortages manifest on a national market remain largely dependent on the national context. Mechanisms such as external price referencing and pharmaceutical expenditure control tools are widely used in the European Union, but their implementation and interaction with other policies (e.g., reimbursement mechanisms, health technology assessment, market structure etc.) differ significantly across countries [70, 94, 95]. As a result, the effects of these policies on drug availability are not uniformly distributed. The available evidence suggest that, in particular contexts, price pressures can lead to launch delays or market withdrawals, while in others the impact is more limited or offset by other factors [95]. In this regard, the situation observed in Romania can be understood as the result of a specific combination of factors (low price levels, dependence on imports, exposure to parallel trade, and a specific regulatory framework with its strengths and weaknesses), which interact in a specific way and amplify the risk of discontinuity. Thus, although the mechanisms identified in this study may reflect trends also present in other European systems, their magnitude and form of manifestation are likely to vary from one country to another [96].

Future research could perform comparative analyses among European Union member states with similar regulatory systems to assess the degree to which the relationships between discontinuities and market withdrawals identified in this study are replicated in other contexts, and to quantify the impact of pricing policy heterogeneity at the European level.

Along these lines, future studies could compare EU member states with market structures different from Romania’s, ranging from markets where prices are higher and where there is local production capacity to those that also rely heavily on external supply sources and/or have low prices. Such comparative perspectives would help us understand whether what we observed in this study also applies to countries less vulnerable to parallel exports, or whether the same dynamics can be seen even where a country is less sensitive to parallel exports or where local production is a primary way of ensuring access to medicines. This could also help a better understanding of the way in which prices, reimbursement policies, and public procurement influence one another and ultimately determine whether medicines reach patients, regardless of the country in which they live.

Policy implications

This study identifies specific policy levers to strengthen the stability of the drug supply in Romania. The upward trend in shortages and the predictable two-year pathway to market exit underscore the necessity of a forward-looking surveillance system capable of preempting supply failures. By monitoring early indicators, including reduced competition for off-patent drugs, unsustainable pricing, and repeated delivery failures, health authorities could employ proactive tools such as specialized procurement or temporary price protections. The particular vulnerability of the generic and biosimilar sectors points to a need for structural reform of the external reference pricing system and the clawback mechanism to ensure that essential, low-cost medicines remain on the market. Furthermore, a more collaborative regulatory environment could help address the systemic pressures that lead to manufacturer exits. These combined efforts would help reorient the health system away from temporary fixes and toward a resilient governance model that prioritizes long-term equity and the reliable delivery of care to all patient populations.

This study suggests that public policies can improve by using ideas from other countries and following Europe’s best practices. Regarding the development of predictive monitoring systems, experience from other jurisdictions shows that it is useful to define quantifiable risk indicators, such as the frequency of discontinuities at the INN level, the degree of market concentration (the number of MAHs for the same INN), or dependence on imports [74, 97–99]. Integrating such indicators into an early warning system or adding them as a feature of each product that falls within these categories (of potential supply disruption) could enable the preventive monitoring of vulnerable products before an actual shortage occurs. In the Romanian context, relevant tools already exist, such as the public service obligation and daily and monthly reporting of stock levels (incoming and outgoing) [85]. However, they function primarily as an ex-post control mechanism, without integration in a dynamic risk assessment system. The latter approach would allow activation of specific interventions (e.g. export restrictions, import facilitation, regulatory flexibilities, stock redistribution etc) before the occurrence of actual disruptions. Regarding pricing policy, a complementary strategy could involve the introduction of elasticity mechanisms for medicines identified as vulnerable; one example could be the generic segment, where price pressures are more pronounced. Such measures could involve targeted price adjustments or changes in reimbursement conditions, accompanied by strict obligations to prioritize availability of those products on the domestic market.

Limitations

This study comes with several limitations to be considered when evaluating its results. First, the interrupted time series analysis used a brief yearly dataset (2017–2024), which reduces statistical strength and the accuracy of estimating slope shifts. Although we used several robustness checks, like negative binomial models and varied error adjustments, small trend alterations post-intervention could go undetected. Therefore, non-significant slope changes, especially post-COVID, warrant careful interpretation and shouldn’t be regarded as definitive evidence that no underlying shifts occurred.

Second, the study relies on administrative and regulatory records, which carry inherent limitations: underreporting, delayed submissions, and evolving notification practices may distort the data. Shortage alerts are dependent on voluntary compliance from manufacturers and distributors, potentially overlooking ground-level supply disruptions at pharmacies or hospitals. Moreover, formal market withdrawals often trail behind actual product unavailability, possibly weakening the observed link between shortages and exits.

A third limitation concerns the study’s scope, which is confined to prescription-only medicines. Over-the-counter products were excluded owing to a lack of standardized pricing and reporting data. Although methodologically justified, this decision limits the applicability of the findings to the wider pharmaceutical market. Furthermore, several potentially relevant variables, such as manufacturing site geography, supply chain concentration, procurement contract specifics, and marketing authorization pathways, could not be included because of their absence from publicly accessible Romanian datasets. While price was used as a core covariate, it relied on the controlled wholesale price from the official pricing structure. It fails to account for non-public, bidding-based pricing, parallel trade influences, or cross-subsidization tactics by manufacturers, all potentially shaping commercial success and discontinuation choices. Consequently, any price effects represent an indicative regulatory price level rather than accurate revenue or profitability values. Finally, the survival analysis, although improved by including time-varying effects and individual INN frailty, is still observational in nature and cannot establish direct cause-and-effect. Unmeasured, confounding factors related to specific products, companies, or therapeutic areas might still be impacting market exit risks. For future research, better predictive models for medicine supply instability could benefit from more comprehensive, international datasets, more frequent updates on shortages, and data on procurement and manufacturing.

An additional limitation stems from the interpretation and application of the discontinuity concept within the national legal framework and from the reliance on the reporting practices of the involved parties. In Romania, the concept of “shortage” is not legally defined by standardized quantitative criteria but is indirectly linked to the obligation of MAHs to report supply disruptions. Therefore, the data used in our study largely reflect administratively reported events, which may be influenced by the need for compliance and the interpretation of legal obligations rather than by a uniform measurement of actual market-level unavailability. This reliance on reporting may introduce additional biases. Some episodes of discontinuity may not be reported at all, either due to delays in notification or to differing interpretations of the threshold at which a case becomes reportable. The legislative framework does set specific deadlines for reporting disruptions, but in practice, of course, there can be time lags between the actual occurrence of a supply problem and its recording in official systems. Therefore, some of our collected data may underestimate or misrepresent the timing of certain episodes of unavailability. Similarly, this approach does not account for situations where existing stocks are insufficient to cover fluctuations in demand or unexpected increases in consumption. Indicators like average monthly consumption, used in current regulatory mechanisms, are not uniformly defined at the national level, for example, by standardizing to actual national consumption, which can lead to differing interpretations of the appropriate supply level among stakeholders. In this context, our results reflect the dynamics of administratively reported discontinuities more than a quantitative measurement of actual access to medicines across the entire market. Although there are nuances in reporting, the strong correlation between official notifications and the well-documented clinical difficulties encountered by healthcare providers indicates that most reported incidents likely reflect actual market-level shortages in most instances.

Another limitation of the study is the lack of data on medication use at the ATC class level, which would have allowed for a more precise contextualization of the observed discontinuities. In the absence of this information, our analysis is based exclusively on the frequency and distribution of the reported discontinuities. We were not, therefore, able to assess the extent to which they are proportional to the level of use of the various therapeutic classes.

While acknowledging these limitations, the convergence of results across diverse analytical approaches validates the strength of our central conclusions, reinforcing the significance of structural factors, rather than incidental supply shocks, as the key drivers of drug shortages and market exit in the Romanian pharmaceutical sector.

Conclusion

This study establishes that pharmaceutical shortages in Romania are a manifestation of systemic market fragility, evidenced by a trend of increasing supply discontinuities that significantly predate the pandemic. Our analysis highlights that periodic shortages often act as a leading indicator of long-term market exit, with risk levels dictated by regulatory and economic constraints rather than immediate logistical shocks. These findings underscore the inadequacy of reactive policymaking and advocate for a paradigm shift toward predictive surveillance and reformed pricing strategies that prioritize market sustainability alongside public health affordability. The integration of several complementary analytical frameworks allows the phenomenon to be investigated from temporal, structural, and dynamic perspectives, contributing to the construction of a coherent methodological framework for assessing discontinuity risks. In this sense, the study highlights the central role of mathematical modeling and statistical analysis in monitoring the pharmaceutical market and informing evidence-based decisions. The proposed approach can serve as a starting point for the development of predictive systems and early warning mechanisms capable of supporting the formulation of proactive public policies aimed at strengthening the resilience of the healthcare system.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contribution

OTC and RA conceptualized the study and drafted the manuscript. RA led the statistical analysis. All authors contributed to data integration, developed the final manuscript structure, reviewed, and approved the final version of the manuscript.

Funding

Publication of this paper was supported by the Carol Davila University of Medicine and Pharmacy, through the institutional program Publish not Perish.

Data availability

All primary data used in this paper are made available at: [https://figshare.com/s/389d0f295a9afce6cfe9]. DOI: 10.6084/m9.figshare.31828048.

Declarations

Ethical approval

Not applicable.

Patient and public involvement

Patients were not involved in the design, conduct, reporting, or dissemination plans of this research.

Competing interests

R.A. received consultancy or speakers’ fees in the past from Merck, UCB, Sandoz, Abbvie, Zentiva, Teva, Laropharm, CEGEDIM, Angelini, Biessen Pharma, Hofigal, AstraZeneca, and Stada. All other authors report no conflicts of interest.