Association Between Newsweek’s Global Hospital Ranking and Patient Outcomes in the USA

INTRODUCTION

Although several hospital rankings have been available in the USA,^{1, 2} there has not been a “global” hospital ranking that assesses hospitals around the world. Recently, Newsweek partnered with a global market research company Statista and ranked the top 1000 hospitals worldwide, including both the USA and foreign hospitals. The ranking was based on three factors: recommendations from peers (55%), patient experience (15%), and quality indicators (e.g., hygiene and patient safety measures, number of patients per doctor/nurse) (30%).³ While informative, this ranking did not account for direct measures of patient outcomes, and it relied heavily on recommendations from peers, despite limited evidence supporting the accuracy of peer evaluations about the quality of hospitals.⁴ To address this issue, we investigated whether the Newsweek’s global hospital ranking of US hospitals was associated with risk-adjusted patient outcomes.

METHODS

We linked three datasets: (1) the Newsweek’s global hospital ranking, (2) the Center for Medicare and Medicaid Services (CMS) Hospital Compare database (for patient outcomes),⁵ and (3) the American Hospital Association Annual Survey database (for hospital characteristics). Patient outcomes were defined as patients’ risk-adjusted 30-day mortality and readmission rates of four major conditions: acute myocardial infarction, heart failure, pneumonia, and chronic obstructive pulmonary diseases (COPD). We calculated composite mortality and readmission rates for each hospital, by calculating the weighted average based on the number of patients with each condition for a given hospital (using 100% Medicare Inpatient File 2015). We restricted to acute care hospitals with 25 or more hospitalizations.

Using this hospital-level-linked dataset, we examined the association between the Newsweek’s global hospital ranking and patient outcomes using multivariate linear regressions with Huber-White heteroscedasticity robust standard errors (Stata 15.1, College Station, TX). We adjusted for hospital size, teaching status, profit status, hospital region, rural/urban status, the presence of ICU, and hospital referral region fixed effects. We weighted the regression models by the total number of admissions for the four conditions for each hospital. The study was approved by the UCLA Institutional Review Board.

RESULTS

Of the 3008 acute care hospitals studied, we found that higher ranked hospitals were more likely to be large, urban, and major teaching hospitals located in the Northeast region (Table 1). After adjusting for hospital characteristics, we found that patients who were treated at the higher ranked hospitals had a lower mortality rate than those treated at the lower ranked hospitals (Table 2). The top 50 hospitals had the lowest risk-adjusted mortality rate of 11.3% (95% CI, 10.9–11.7%), followed by 11.9% (95% CI, 11.5–12.2%) for hospitals ranked 51–100, 12.1% (95% CI, 11.8–12.3%) for 101–150, 12.1% (95% CI, 11.8–12.5%) for 151–200, and 12.4% (95% CI, 12.3–12.4%) for hospitals ranked 201 or lower (P-for-trend < 0.001). Although the association was somewhat weaker, we found a similar pattern for patients’ readmission rate (P-for-trend = 0.004).

Table 1 Characteristics of Hospitals by Newsweek’s Hospital Ranking Category

Table 2 Adjusted Association Between the Rank Categories and Patient’s Outcomes

DISCUSSION

Using a nationally representative sample of acute care hospitals in the USA, we found that patients who were treated at the high-ranked hospitals in the Newsweek’s ranking had lower mortality and readmission rates compared with those treated at the lower ranked hospitals. These findings support the validity of Newsweek’s global hospital ranking, developed based on peer recommendations, patient experience, and the structural and process-of-care measures of quality.

Our study has limitations. First, although we adjusted for a broad set of hospital characteristics, we could not preclude the possibility of the residual confounding. Second, our sample included only hospitals in the USA, and therefore, Newsweek’s hospital rankings may not be associated with patient outcomes in other countries. Finally, since our analysis was based on the outcomes of Medicare beneficiaries, these findings may not be generalizable to a younger population.