Table 2.
Studies that compare patient online reviews with traditional healthcare quality indicators.
| Study | Comparator measures (patient surveys, clinical outcomes, or other quality measures) | Comparison methods and results |
| Greaves et al, 2012 [33] | (1) Mailed-based patient surveys. (2) Clinical outcomes from the National Health Service (NHS) Information Center and NHS Comparators (eg, The proportion of patients with diabetes receiving flu vaccinations, proportion of hypertensive patients with controlled blood pressure, proportion of diabetic patients with controlled HbA1C, percentage of low-cost statin prescribing, cervical screening rate, admission rates for ambulatory care sensitive conditions, and the proportion of achieved clinical Quality and Outcomes Framework (QOF) points from available points. N (POR)=16,592, N (physicians)=4934. | (1) ρ =0.37~0.48, P<.001 for Pearson correlation of POR and survey. (2) ρ=–0.18~0.18, P<.001 for the correlation of POR and clinical outcomes. |
| Greaves et al, 2012 [34] | Traditional survey of patient experience. N (POR)=9,9997, N (physicians)=146. | ρ=0.13~0.49, P<.001 for Pearson correlation of POR and survey. |
| Segal et al, 2012 [59] | Volume of surgeries. N of POR=588, N of surgeons=600. | High volume surgeons have higher mean values of PORs than low-volume surgeons, but effect size was weak. |
| Bardach et al, 2013 [53] | (1) Overall hospital ratings on HCAHPS. (2) Hospital individual HCAHPS domain scores (eg, nurse communication, pain control). (3) Hospital 30-day mortality and hospital 30-day readmission rates. N (POR)=3796, N (hospitals)=962. | Pearson correlation (n=270), ρ=0.49, P<.001 for 3 out of 4 measures. Higher ratings were associated with lower mortality and readmission rates. |
| Wallace et al, 2014 [11] | (1) Likelihood of patient visiting their primary care physician within 14 days of hospital discharge. (2) Health care expenditure. N (POR)=58,110, N (physicians)=19,636. | (1) Regression model for sentiment generated from POR comments and the comparator r2=.21, P=.03; (2) Regression model for POR rating combined with topics generated from POR comments r2=.25. |
| Glover et al, 2015 [52] | 30-day hospital-wide all-cause unplanned readmission rate (HWR). POR=Facebook comments. POR=Facebook comments, N (hospitals)=136. | Independent sample t test (n=315 vs 364), POR=4.15±0.31 vs 4.05±0.41, P<.01 more PORs was associated lower HWR. |
| Emmert et al, 2015 [31] | (1) Quality measures on cost of medication, type 2 diabetes-related intermediate outcome measure, and patient/doctor ratio from German Integrated Health Care Network (QuE); (2) German patient satisfaction survey from QuE. N (POR)=1179 on Jameda, N=991 on Weisse Liste. N (physicians)=69. | (1) Spearman’s rank correlation (n=991) ρ=0.297~.384, P<.05 for cost per prescription; ρ=0.478, P<.05 for patient with HbA1c-target values; ρ=−0.316~−0.289, P<.05 for patient/doctor ratio on Weisse Liste; (n=1179) ρ=0.298, P<.05 for cost per case, ρ=0.298~386, P<.05 for patient/doctor ratio on Jameda; (2) Spearman’s rank correlation (n=991), ρ=−0.347~−0.372, P<.05 for 3 out of 4 measures on Weisse Liste; (n=1179), ρ=−0.391~0.640, P<.05 for all measures on Jameda. |
| Okike et al, 2016 [56] | Risk-adjusted mortality rate. N of POR NAa, N (surgeons)=590. | Pearson’s correlation (n=590), r=−.06, P=.13. |
| Bardach et al, 2016 [51] | Researchers identified HCAHPS domains. N (POR)=244 (narratives), N (hospitals)=193. | Content analysis (139/244, 57% of POR comments mentioned HCAHPS domains). |
| Kilaru et al, 2016 [47] | HCAHPS inpatient care surveys. N (POR)=1736, N (Emergency Departments)=100. | Content analysis. Considerable overlaps in theme of PORs and HCAHPS domains. |
| Ranard et al, 2016 [50] | Researchers identified HCAHPS domains. N (POR)=16,862, N (hospitals)=1352. | Content analysis. POR comments covered 7/11 HCAHPS domains and introduced 12 new domains not existing in HCAHPS. |
| Emmert et al, 2018 [44] | Hospital-level quality measures by the CMS. N (POR)=1000, N (hospitals)=623. | (1) Spearman’s correlation ρ=±0.143, P<.05 for 13 of 29 measures; (2) Spearman’s correlation ρ=±0.114, P<0.05 for 7 of 29 measures, indicating weak association. |
| Trehan et al, 2018 [68] | Total knee replacement (TKR) outcomes: infection rate, 30-day readmission rate, 90-day readmission rate, revision surgery. N of POR NA, N (surgeons)=174. | Kruskal–Wallis one-way analysis of variance one-way analysis of variance (one-way ANOVA on ranks) showed no correlation. |
| Campbell et al, 2018 [66] | 1) HCAHPS patient satisfaction measures; 2) HCAHPS hospital-wide 30-day readmission rate; 3) Medicare spending per beneficiary ratio. N of POR NA, N (hospitals)=136. | (1) Multivariable linear regression (n=136), r2=.16~.5, P<.05 for 21 of 23 measures; Pearson’s correlation (n=136), r=.27~.61, P<.005 for 19 of 23 measures; (2) Multivariable linear regression r2=−.58, P<.10 for readmission rate; (3) Multivariable linear regression r2=−.006, P<.731 for Medicare spending per beneficiary. Overall weak association. |
| Jarari et al, 2018 [71] | Nursing Home Compare (NHC) website quality measures. | POR rating was significantly different from NHC rating. |
| Chen et al, 2018 [67] | Press Ganey Medical Practice Survey for patient satisfaction. N of POR NA, N (physicians)=200. | Pearson’s correlation (n=226), r=.18, P<.001. |
| Daskivich et al, 2018 [69] | Specialty-specific performance scores (adherence to Choosing Wisely measures, 30-day readmissions, length of stay, and adjusted cost of care), primary care physician peer-review scores, and administrator peer-review scores. | Multivariable linear regression (n=30) r=−.04, P=.04. |
aNA: not available.