Patient Safety in Family Medicine through the Eyes of People with Chronic Conditions in Slovenia

This cross-sectional study was part of the international Patient-Reported Indicator Surveys (PaRIS) programme, coordinated by the Organization for Economic Co-operation and Development (OECD) (23). The Slovenian part of the study was conducted in 2023 at 70 randomly selected family medicine practices, representing various geographic and demographic areas (28,29,30). The OECD developed the overall PaRIS survey design and conceptual framework in consultation with several international collaborators, including the Netherlands Institute for Health Services Research (NIVEL), Ipsos MORI, the University of Exeter, the Avedis Donabedian Institute, and OptiMedis AG (23).

Two validated OECD PaRIS instruments were used.

At the practice level, the Primary Care Practice Questionnaire (PaRIS-PCPQ) was administered. The full instrument contains 40 items covering organisational characteristics, workforce composition, care coordination, chronic care management, and the use of electronic health information systems (31). For the purposes of this analysis, the following PCPQ items were used:

• (1)

  practice location,

• (2)

  practice type,

• (3)

  electronic exchange of patient information with external providers,

• (4)

  review of indicators of patient care, and

• (5)

  routine development of patient care plans for groups of patients.

At the patient level, the Primary Care Patient Questionnaire (PaRIS-PQ) was used. The full instrument comprises 121 items addressing sociodemographic characteristics, health behaviours, health status, patient-reported outcomes, and patient experiences of care (32). In the present study, 7 patient-safety-related items from the PaRIS-PQ (Q69–Q75) were analysed, covering:

• number of medications taken regularly,

• medication review in the previous 12 months,

• repetition of medical information,

• reported adverse events,

• whether respondents were encouraged to express concerns,

• foregoing care due to transport difficulties, and

• foregoing care due to medical costs.

Both instruments were translated into Slovenian and culturally adapted using the TRAP-D process (Translation, Review, Adjudication, Pretesting, and Documentation) to ensure linguistic accuracy and conceptual equivalence (29).

Data collection took place from April to June 2023. Participation was voluntary and anonymous. The study adhered to the General Data Protection Regulation (EU) 2016/679. All responses were anonymised before analysis, and no personally identifiable data were collected or stored.

The primary outcome variable was reported adverse events derived from PaRIS-PQ item Q72. This item asked respondents how often they had experienced harmful or adverse events related to their care. For regression analysis, this variable was dichotomised into the presence of adverse events versus no adverse events, as specified below.

Independent variables at the patient level included:

• number of regularly taken medications,

• medication review within the previous 12 months,

• repetition of medical information to healthcare providers,

• lack of encouragement for respondents to express their concerns,

• barriers to care due to transport difficulties, and

• barriers to care due to medical costs.

• At the provider level, independent variables included:

• practice location (urban/rural),

• practice type (public/private concession practice), and

• electronic exchange of patient information with external healthcare providers.

Descriptive statistics were used to summarise provider-and patient-level characteristics, including frequencies, percentages, means, and standard deviations, as appropriate. Associations between categorical variables were examined using chi-square tests, and associations between ordinal patient-safety indicators were assessed using Spearman’s rank correlation coefficients.

The primary outcome for multivariable analysis was reported adverse events (Q72). For binary logistic regression, responses were dichotomised as follows: respondents reporting adverse events “always”, “often”, or “sometimes” were classified as having experienced an adverse event, whereas those responding “rarely” or “never” were classified as not having experienced an adverse event. Responses marked as “not suitable for me” or missing were excluded from the regression model.

Binary logistic regression was used to identify predictors of reported adverse events. Independent variables included the number of regularly taken medications, medication review within the previous 12 months, repetition of medical information, lack of encouragement for respondents to express concerns, barriers to care due to transport difficulties, and barriers to care due to medical costs. Results are presented as regression coefficients (B), standard errors (SE), Wald chi-square statistics, p-values, odds ratios (ORs), and 95% confidence intervals (CIs).

Because respondents were clustered within practices, a multilevel mixed-effects logistic regression model with a random intercept for practice was applied in accordance with OECD PaRIS analytic procedures. Model fit was assessed using the model chi-square statistic and Nagelkerke R². Statistical significance was set at p < 0.05. Analyses were performed in SPSS version 31.

The study was conducted in accordance with the principles of the Declaration of Helsinki and approved by the National Medical Ethics Committee of the Republic of Slovenia (No. 0120-260/2021/3). All participants provided informed consent before inclusion in the study.

Table 3 summarises the responses from respondents regarding patient safety indicators, including the pattern of medication use, medication review by healthcare professionals, communication safety, adverse events, encouragement of respondents to express concerns, and access barriers such as transport and cost.

Cross-tabulation analyses showed statistically significant associations between medication burden and several patient safety indicators. Respondents taking a higher number of medications were more likely to report having had a medication review, to report repeating their medical information, and to report adverse events. They were also more likely to report barriers to care related to transport and costs (all p < 0.001). These findings suggest that respondents with a greater medication burden may also face greater safety-related vulnerability.

Spearman’s correlation analyses showed that repetition of medical information was positively associated with reported adverse events (r_s = 0.16, p < 0.001). Transport barriers (r_s = 0.12, p < 0.001) and cost barriers (r_s = 0.15, p < 0.001) were also positively associated with adverse events. Encouragement to express concerns was negatively associated with adverse events (r_s = −0.08, p = 0.004). Although these associations were statistically significant, their magnitude was weak and should therefore be interpreted with caution (Table 5).

A multivariable logistic regression analysis was conducted to identify predictors of reported adverse events. The overall regression model was statistically significant (χ²(6) = 24.8, p < 0.001), although the proportion of explained variance was low (Nagelkerke R² = 0.02) (Table 6).

Among the examined predictors, repetition of medical information was significantly associated with a higher likelihood of reporting adverse events (B = 0.42, SE = 0.17, Wald χ² = 6.13, p < 0.001). Respondents who reported frequently having to repeat their medical information had approximately 1.53 times higher odds of reporting adverse events compared with those who did not (OR = 1.53, 95% CI 1.20–1.95). Similarly, lack of encouragement to express concerns was associated with increased odds of adverse events (B = 0.76, SE = 0.47, Wald χ² = 2.65, p = 0.008). Respondents who reported less encouragement to express concerns had 2.14 times higher odds of reporting adverse events (OR = 2.14, 95% CI 1.14–4.03). Transport barriers to care were also a significant predictor (B = 0.32, SE = 0.21, Wald χ² = 2.26, p = 0.024), indicating that respondents facing transportation difficulties had slightly higher odds of reporting adverse events (OR = 1.10, 95% CI 1.02–1.95). In contrast, the number of medications, medication review within the previous 12 months, and cost-related barriers to care were not statistically significant predictors in the adjusted model (all p > 0.05). For example, the number of medications taken regularly showed no significant association with adverse events (B = −0.28, SE = 0.34, p = 0.492; OR = 0.75, 95% CI 0.34–1.62).

Medication-related safety remains a key concern in primary care, especially among people with chronic conditions and multimorbidity. In our study, a substantial proportion of respondents reported regular medication use, yet only a minority reported having undergone a medication review within the previous 12 months. This finding suggests potential gaps in systematic medication monitoring in family medicine.

Polypharmacy is widely recognised as an important risk factor for adverse drug events, hospitalisations, and inappropriate prescribing, particularly in older populations. Numerous studies have shown that increasing medication burden is associated with a higher risk of medication-related harm (1–2, 4, 10). However, in the present study polypharmacy was not significantly associated with reported adverse events in the adjusted regression model. This suggests that medication burden alone may not fully explain safety outcomes in this dataset. Instead, the relationship between polypharmacy and patient safety may be mediated by other factors, such as the quality of medication review, care coordination, and patient-provider communication.

These findings are consistent with previous research suggesting that medication-related safety risks often arise not only from the number of medications but also from system-level factors such as inadequate medication reconciliation and fragmented care processes (6, 18–19).

One of the strongest predictors of reported adverse events in this study was the need for respondents to repeat medical information that should already be available in the medical record. This finding likely reflects fragmentation of health information systems and limited interoperability between providers. When patient information is not consistently accessible across care settings, the risk of communication errors, duplicated tests, and inappropriate treatment decisions may increase.

Similarly, respondents who reported less encouragement to express their concerns had higher odds of reporting adverse events. This finding highlights the importance of patient engagement and open communication in maintaining safe care processes. Previous studies have shown that when people feel comfortable voicing concerns and participating in decision-making, healthcare teams are more likely to detect potential safety issues early and prevent harm (27, 33–34).

Despite this, a large proportion of respondents reported that they were rarely or never encouraged to express concerns during consultations. This finding suggests that patient empowerment and shared decision-making may still represent an underdeveloped aspect of patient safety culture in primary care.

Transport barriers were also associated with a higher likelihood of reporting adverse events. Although Slovenia has a largely publicly funded healthcare system, physical access to care may still represent an important determinant of patient safety, particularly for older adults and people with multimorbidity.

Barriers to transport may lead to delayed consultations, missed follow-up appointments, and reduced continuity of care, all of which may increase the risk of adverse outcomes. By contrast, cost-related barriers were not statistically significant predictors in the adjusted model, suggesting that in the Slovenian context physical accessibility may play a more important role than financial barriers in shaping patient safety experiences.

Approximately one in six respondents reported experiencing an adverse event at least occasionally. These results should be interpreted cautiously, as the survey item captures patient-reported experiences rather than clinically validated safety incidents. Patient-reported safety indicators provide valuable insight into perceived problems in care processes, but they may reflect a broader range of experiences than formally documented medical errors.

Nevertheless, respondent perspectives are increasingly recognised as an important component of safety monitoring, particularly in primary care where many safety events remain undocumented in clinical reporting systems (27, 33–34).

The regression model explained only a small proportion of the variance in reported adverse events (Nagelkerke R² = 0.02). Although this value appears low, such findings are common in studies examining complex social and behavioural phenomena such as patient safety in primary care. Safety outcomes are influenced by multiple interacting determinants at the patient, provider, organisational, and health-system levels, many of which were not captured in the present analysis.

Similarly, the correlations observed between safety indicators were generally weak. This pattern suggests that patient safety incidents in primary care are multifactorial and cannot be explained by individual indicators alone. Even so, identifying statistically significant predictors can still provide useful insights into areas where improvements in care processes may reduce safety risks.

This study has several strengths. It uses data from the OECD PaRIS programme, applies internationally developed and culturally adapted instruments, and includes a large national sample of Slovenian people with chronic conditions in family medicine.

The study also has important limitations. First, the cross-sectional design does not allow causal inference. Second, all patient-level data were self-reported, meaning that the adverse events captured in this study reflect respondent’ perceptions and experiences rather than clinically validated safety incidents. Third, the dichotomisation of the adverse-event item may have reduced nuance by combining different frequencies of reported events into a binary outcome. Fourth, the observed correlations were weak, and the regression model explained only a small proportion of the variance, indicating limited predictive power. Fifth, participation was voluntary, which raises the possibility of non-response bias, including possible overrepresentation of more engaged practices or respondents with higher health literacy. These factors may limit generalisability.

Despite these limitations, the findings point to several actionable areas for improvement. First, medication review should be strengthened in routine family medicine care, especially for people with multimorbidity and higher medication burden. Second, digital interoperability and information continuity should be improved to reduce the need for respondents to repeat medical information across providers. Third, communication practices should more consistently support people in expressing concerns, as this may contribute to safer and more person-centred care. Finally, access barriers—particularly transport-related barriers—should be considered within national patient-safety strategies, because safe care depends not only on clinical decisions but also on people’s ability to reach and navigate services.

Regular medication review, better digital information exchange, and incorporation of patient-reported experience and outcome measures into routine quality monitoring may represent feasible system-level steps toward improving patient safety in Slovenian family medicine.