The Hidden Variable in Hospital Digitalization: Landmark Study Finds Software Vendor Choice Shapes Digital Maturity Across 1,600 Hospitals
When a hospital signs a contract with a health information technology vendor, the decision typically stays confined to the boardroom and the information technology department. A sweeping new study suggests it may deserve a place among the most consequential choices a hospital ever makes. Drawing on data from more than 1,600 German hospitals — a near-complete census covering roughly 90 percent of all hospitals in the country — researchers found that the characteristics of a hospital’s Hospital Information System, or HIS, provider are significantly associated with how digitally mature the institution is, even after statistically accounting for hospital size, ownership, and teaching status. The study, published in the Journal of Medical Systems by Jonas Backes, Alexander Geissler, Jonas Subelack, and David Ehlig, draws on the DigitalRadar project, one of the most comprehensive surveys of hospital digitalization ever assembled, and offers some of the first systematic evidence that the structure of the software market itself may be sorting hospitals into digital haves and have-nots.
Hospitals worldwide are pouring resources into digital infrastructure with three explicit goals: greater efficiency, greater safety, and more patient-centered care. At the center of virtually all of these efforts sits the Hospital Information System — the interconnected family of software that stores electronic patient records, routes laboratory results, manages medication orders, schedules operating theaters, coordinates discharge, and handles billing. When such a system works well, a physician’s order flows straight to the pharmacy and a discharge summary reaches the family doctor without a fax machine in sight. It is, in effect, the digital nervous system of a modern hospital. What makes the new research striking is the nature of the market supplying these systems. The HIS provider landscape is fragmented and heterogeneous, populated by vendors that differ enormously in the breadth of their software portfolios, the degree to which their modules communicate with one another and with external partners, and the size of the hospitals they are built to serve. Whether that underlying market structure actually influences a hospital’s capacity for digital transformation has, until now, been remarkably underexplored.
To close that evidence gap, the research team turned to the DigitalRadar dataset, a German-wide project that records 234 standardized data items for every participating hospital and assigns each institution a digital maturity score ranging from 0 to 100. Because the survey reaches about nine in ten German hospitals, it behaves less like a conventional sample and more like a national census, giving researchers an unusually complete picture of an entire hospital landscape. Crucially, the investigators shifted the unit of analysis away from hospitals and onto the software vendors themselves. From the data they derived four characteristics describing each HIS provider: module utilization, a measure of how extensively hospitals actually deploy the software modules a vendor offers; integration ratio, which captures how thoroughly a provider’s systems are connected; external provider variation, a measure of how differently a vendor’s client hospitals combine outside information technology services; and maximum hospital size coverage, which reflects the largest institutions a provider serves.
With those four features in hand, the researchers applied k-means clustering, a workhorse machine-learning technique that partitions observations into a predefined number of groups by minimizing the distance between each data point and the center of its cluster. The persistent challenge in such analyses is deciding how many clusters to allow, so the team tested every solution from two groups to ten and judged each with two standard diagnostics. The first, the within-cluster sum of squares, measures how tightly providers sit inside their groups; plotted against the number of clusters, it declined steeply up to four clusters, with successive reductions of 50.0, 14.8, 9.8, and 4.1 as the solution moved from two to five clusters — a drop that more than halved between four and five and flattened thereafter, the classic elbow that signals diminishing returns from adding groups. The second diagnostic, the silhouette score, quantifies how well each provider fits its own cluster relative to neighboring ones; it peaked at four clusters with a value of 0.476 — technically tied with the two-cluster solution — exceeding the 0.454 recorded at three clusters and the 0.398 recorded at five. Because the four-cluster solution, unlike the two-cluster alternative, preserved meaningful heterogeneity among providers, satisfied the minimum cluster size criterion, and showed acceptable bootstrap stability, the researchers settled on four groups as the natural architecture of the German HIS market.
The clustering revealed four distinct species of vendor. Cluster A, comprising six providers, combined high module utilization with high integration and was typically deployed in large hospitals; it was also by far the most common configuration in the country, covering 1,301 hospitals. Cluster B contained three providers showing similarly high module utilization but limited external connectivity — powerful systems that reach less readily beyond the hospital walls. Cluster C, with four providers, exhibited very low utilization and integration ratios and served predominantly small or psychiatric hospitals, a niche configuration that touched just seven hospitals in the entire dataset, a sample so small that the researchers explicitly urge cautious interpretation of any statistics attached to it. Cluster D was the largest group of vendors, eleven providers offering moderate utilization and integration mainly to small- and mid-sized hospitals. In total, the clustering mapped 24 providers serving more than 1,600 institutions, from deeply integrated enterprise platforms to minimally digitalized niche suppliers.
Beneath the taxonomy lies a stark portrait of market concentration. The analysis found only eight providers active in hospitals with more than 700 beds, and a mere three providers operating in hospitals with more than 2,000 beds — a funnel that narrows dramatically as institutions grow larger and their information technology needs become more complex. Cluster A dominated across every size category, capturing its highest share, 87 percent, among mid-to-large hospitals, while Clusters C and D were concentrated among smaller institutions. The pattern inverts what one might hope for in a healthy digital ecosystem: the largest hospitals, which command the greatest resources and technical staffing, choose from a tiny pool of large-scale vendors, while smaller hospitals — often those with the least capital and the thinnest IT departments — navigate a more crowded field of providers whose systems are less fully used and less completely connected. Size, in other words, buys not only technology but access to the technology that matters.
The statistical heart of the study lies in its multivariate linear regressions, which quantified the association between provider characteristics and digital maturity while controlling for hospital characteristics. Higher module utilization was associated with significantly higher digital maturity, with a regression coefficient of 6.87 — a substantial effect on a scale that runs only from 0 to 100. At the cluster level, hospitals running systems from Cluster B providers consistently showed the highest digital maturity, with the advantage concentrated in the Structures and Systems domain and in administrative workflows. Hospitals served by Cluster D providers scored lower overall, with the deficit most pronounced in telehealth, while both Clusters C and D exhibited weak digital maturity in discharge workflows — the processes that move patients out of the hospital and back into community care. Clusters A and B, meanwhile, were concentrated among larger hospitals.
How should the pattern be read? The most intuitive interpretation is that what a vendor offers — and how much of it hospitals actually put to work — helps set a ceiling on what an institution can achieve digitally. A hospital cannot be more integrated than its core systems allow, and it cannot automate workflows that its software does not meaningfully support. The maturity score measures achievement, and achievement depends on money, staffing, leadership, and strategy as much as on software. Yet the researchers are careful about causal claims. The study is observational, and a regression that controls for hospital characteristics cannot fully rule out reverse causality: digitally ambitious hospitals may simply select particular vendors, or negotiate richer deployments once they adopt them. The seven hospitals in Cluster C are far too few to support firm conclusions, a limitation the authors acknowledge directly. Still, the fact that the associations survived adjustment for hospital size, ownership, and teaching status suggests that vendor characteristics carry information about digital maturity that hospital structure alone cannot explain.
The implications ripple well beyond Germany’s borders. For hospital executives, the findings elevate procurement from an IT formality to a strategic decision with measurable consequences: choosing a system means choosing a trajectory, because the depth of module use and the degree of integration appear to travel with maturity scores. For policymakers, the study adds a new variable to the debate over healthcare digitalization — the structure of the software market itself. If a handful of providers dominate large hospitals while smaller institutions depend on less utilized, less integrated systems, then inequalities in patient-facing digital services such as telehealth and digital discharge management may originate partly in the vendor landscape rather than in clinical ambition or funding alone. For patients, the stakes are concrete: telehealth availability and smooth discharge processes — both domains where weaker vendor clusters lagged — shape how easily people reach care remotely and how safely they leave the hospital. Germany, now equipped with census-level measurements through DigitalRadar, offers other national health systems a template for auditing whether their own software markets are broad enough, deep enough, and connected enough to support genuine transformation.
The study is published open access, with detailed results for each HIS variable and each anonymized provider supplied in the supplementary materials, an unusual degree of transparency for research that touches a commercial market. As DigitalRadar continues to measure German hospitals, future analyses can begin to disentangle whether vendor characteristics drive digital maturity or merely travel alongside it — a question the current cross-sectional design leaves open. For now, the message to hospital boards is blunt. Digital transformation is not simply a matter of buying technology; it is a matter of how completely that technology is deployed, how thoroughly it is connected, and how deeply it penetrates daily workflows. And in a market where just three vendors serve every German hospital above 2,000 beds, the software contract may indeed be among the most consequential documents a hospital ever signs.
Subject of Research: The relationship between Hospital Information System provider characteristics and digital maturity across more than 1,600 German hospitals, covering approximately 90 percent of all hospitals nationwide.
Subject of Research: Medicine
Article Title: Does System Choice Matter? Influence of Hospital Information Systems on Digital Maturity
Article References: Backes, J., Geissler, A., Subelack, J., & Ehlig, D. (2026). Does System Choice Matter? Influence of Hospital Information Systems on Digital Maturity. Journal of Medical Systems, 50(1), Article 124. https://doi.org/10.1007/s10916-026-02450-w
Image Credits: AI Generated
DOI: 10.1007/s10916-026-02450-w
Keywords: Hospital Information Systems; Digital Maturity; Hospital Digitalization; Health Information Technology; K-means Clustering; DigitalRadar; Market Concentration; Telehealth
Cite Scienmag News
APA MLA Chicago
Ophelia Keating. (August 30, 2026). Hospital Information System Choice Shapes Digital Maturity, Study Finds. Scienmag. https://scienmag.com/hospital-information-system-choice-shapes-digital-maturity-study-finds/
Ophelia Keating. “Hospital Information System Choice Shapes Digital Maturity, Study Finds.” Scienmag, 30 August 2026, https://scienmag.com/hospital-information-system-choice-shapes-digital-maturity-study-finds/. Accessed 30 August 2026.
Ophelia Keating. “Hospital Information System Choice Shapes Digital Maturity, Study Finds.” Scienmag. August 30, 2026. https://scienmag.com/hospital-information-system-choice-shapes-digital-maturity-study-finds/
Copy citation Download RIS
Tags: determinants of hospital digitalizationdigital maturity in hospitalsdigital transformation in healthcareGerman hospital digitalization studyHealth Information Systems effectivenesshealth information technology vendor selection effectshealthcare digital infrastructure developmenthealthcare software market influencehospital digital infrastructure developmenthospital digital transformationhospital digitalization barriers and facilitatorshospital information system selectionhospital information technology decision-makinghospital size and digital maturityhospital size and ownership influence on digital maturityhospital software market influenceimpact of HIS vendor choiceimpact of HIS vendors on hospital digitalizationrole of HIS provider characteristicsrole of software vendors in healthcare
