Using linked data from the Norwegian Hip Fracture Register (NHFR) and Statistics Norway (SN), we performed a nationwide (5.3 million inhabitants at the end of the study period in 2018) retrospective cohort study of prospectively collected data.

Data on hip fracture patients (ICD-10 [14] codes S72.0-S72.2) operated on in Norwegian hospitals have been collected by the NHFR since 2005 [15], and baseline patient characteristics (sex, age, American Society of Anesthesiologists physical status (ASA) class [16], type of surgery given and presence of chronic cognitive impairment) were extracted. Hip fractures treated with total hip arthroplasty were recorded in the Norwegian Arthroplasty Register and subsequently imported to the NHFR. Type of treatment was categorized into Osteosynthesis (i.e. cannulated screws, dynamic hip screws, Intramedullary nails), Hemiartrhoplasty, Total hip arthroplasty and Other (combined interventions or rare treatment methods).

The date of death was imported to the NHFR from the National Population Register. Patient Reported Outcome Measure (PROM) questionnaires were sent from the NHFR to all living patients at 4, 12, and 36 months postoperatively. No reminders were sent to non-responders. Patients treated with THA (n = 1,694) only completed the four-month PROM questionnaire.

The PROM questionnaire included three measures of health, all of which used a visual analog scale (VAS). First, the EQ-VAS is based on respondents’ direct valuations of their overall health-related quality of life (HRQoL) on a scale that ranges from 0 (worst imaginable health) to 100 (best imaginable health) [17]. Second, the Pain-VAS measuring the patients´ average self-reported pain from the operated hip the last month before the defined follow-ups, ranging from 0 to 100 (100 represents the worst possible pain). Third, the Satisfaction-VAS measuring satisfaction with the result after the surgical treatment of the fractured hip and is reported on a scale from 0 to 100, where 0 represents the most satisfactory score. For analytical purposes and presentation in the paper, the scale was inverted, i.e., 100 represents most satisfied, and 0 represents least satisfied. Note that the Satisfaction-VAS explicitly refers to ‘the result after the surgical treatment’, i.e. its outcome. We therefore categorized it as a PROM-variable, rather than a PREM that refers to experience with the treatment per se.

The completeness of reporting in the NHFR has been regularly evaluated and was 88% for osteosynthesis, 95% for hemiarthroplasties, and 88% for total hip arthroplasty in 2015-2016 [18]. This completeness evaluation was performed in the middle of the inclusion period, and we assume that these figures are valid for the data used in this study.

We acquired individual socioeconomic position data from Statistics Norway (SN), including household income and the highest level of education attained. Household income, defined as the household’s total taxable income (including wages, social benefits, pensions, etc.) in the year prior to the injury, was divided into three equally sized groups: low (< 20,500 Euro), medium (20,500 − 36,600 Euro), and high (> 36,600 Euro-). Converted from Norwegian crowns (NOK) January 2025. The International Standard of Classification of Education classified educational levels into three levels [19]: low, which represents lower secondary education; medium, which represents upper secondary to short-cycle postsecondary education; and high, which represents bachelor’s degree and higher education.

As of 31 December 2019, the NHFR included data on 41,699 fractures with a minimum of one year of follow-up from 1 January 2014 to 31 December 2018. Since it would have been challenging to maintain distinct follow-up periods for fractures occurring on both sides, patients with bilateral fractures were excluded during the inclusion period (n = 4,018 in 2,009 patients). These patients were of slightly higher age and exhibited greater comorbidity, as measured by ASA classification and the presence of cognitive impairment, compared to the study population. Furthermore, patients with pathological fractures (n = 400), patients with missing data on ASA class (n = 402), and patients with missing data in the coupled datasets (NHFR and SN) (n = 1,673) were excluded (Fig. 1). Thus, data from 35,206 patients were available for analysis.

Patient inclusions and exclusions

Legend: Norwegian Hip Fracture Register (NFHR) - Statistics Norway (SN) - American Society of Anesthesiologists physical status class (ASA)

Demographic variables are presented as absolute numbers and percentages. The time trend data are crude mean values stratified separately for the 3 categories of education and of household income. The level of significance was set at p < 0.05 in all analyses.

To provide a simple visual representation of the potential causal relationships among the variables included in this study and to select variables to include in the statistical analyses for adjustment, we applied a directed acyclic graph (DAG) model (http://www.daggity.net) (see Fig. 2). The DAG representation is useful for determining whether a given pair of variables are independently associated and the empirical directions of effects. The DAG representation makes explicit what assumptions are being made if causal relationships are not ascertained and is particularly useful when estimating causal relationships from non-randomized studies, which are subject to confounding factors.

Directed acyclic graph (DAG) visualizing possible causal relationships among available covariates

Legend: All included covariates presented. Socioeconomic position is represented by level of education and household income. ASA class - American Society of Anesthesiologists physical status represents observed comorbidity at surgery

To assess the associations between SEP (level of education and level of household income) and the mean VAS score on the three VAS scales (EQ-VAS, Pain-VAS, and Satisfaction-VAS), we performed an analysis of covariance (ANCOVA) utilizing a generalized linear model (GLM) with the PROC GLM function in SAS Statistics. We performed six separate analyses at all three follow-ups: categories of education and household income versus EQ-VAS, Pain-VAS, and Satisfaction-VAS scores. To assess the magnitude of difference in mean VAS scores between categories of education and household income on the EQ-VAS, Pain-VAS, and Satisfaction-VAS scores at 4, 12, and 36 months, the same analysis of covariance (ANCOVA) as described before was performed. The magnitudes and confidence intervals (CIs) of estimations were calculated. Cohen´s d as measure of effect size was estimated as mean difference in the separate VAS scores divided by the mean square error. All analyses were adjusted for age, sex, ASA class, type of surgery, and presence of chronic cognitive impairment. In addition, all analyses were adjusted for the SEP factor that was not utilized as independent variable. I.e. analyses on education were adjusted for income and vice versa.

The analyses were performed using SAS/STATS for Windows v. 8.3 (SAS Institute, Cary, North Carolina, USA). The STrengthening the Reporting of OBservational Studies in Epidemiology (STROBE) guidelines were followed [20].

The Northern Norway Regional Committee for Medical and Health Research Ethics approved the project and exempted it from the duty of confidentiality (REK 2018/1955). A data integrity assessment was conducted in accordance with the EU General Data Protection Regulation (GDPR). The project was funded by the Northern Norway Regional Health Authority (HNF1482-19). The NHFR is financed by the Western Norway Regional Health Authority. There are no competing interests to declare.