Predictable and less predictable undesirable effects of improvement interventions in healthcare
| Predictable undesirable effects (‘balancing measures’)13 | |
| Efforts to reduce hospital length of stay might increase short-term readmissions. | In their efforts to lower length of stay, clinicians might discharge patients prematurely. |
| Performance measures can lead to inappropriate clinical care, decreased provider attention to other patient concerns.52 | Labelling patients with minimal/early evidence of chronic condition makes it easy to meet targets for disease control. Clinicians ‘play to the test’ and focus primarily on the few aspects of care being measured. |
| Making 'time to antibiotics' a quality target led to overdiagnosis and overtreatment of pneumonia. | The measure encouraged clinicians to label patients with pneumonia and administer antibiotics right away rather than taking time to sort out the diagnosis and risk missing target treatment time.14 15 |
| Quality measures focused on inpatient falls might decrease mobility for older patients.53 54 | Penalising hospitals for falls (even non-injurious ones) may reduce efforts by nurses and physiotherapists to mobilise older patients in hospital. |
| Copayments may reduce medication adherence. | Poorer patients may not fill prescriptions even for necessary medications.55 |
| Prescribing alerts with (nearly) hard stops can cause treatment delays. | A trial of an alert focused on the interaction between warfarin and trimethoprim-sulfamethoxazole was terminated after several patients experienced potentially serious treatment delays.23 |
| Less predictable undesirable effects (‘Unintended Consequences’) | |
| Temporary increase in mortality in paediatric ICU after introducing CPOE—probably due to delayed ordering and administration of urgent medications.22 | Probably due to delayed administration of urgent medications due to cumbersome order entry as well as problems entering ordering for patients who are critically ill and are not yet registered in the system.22 |
| Wrong patient orders likely occur more frequently with CPOE compared with paper charts.56 | Probably various human factors such as similarity of screens for all patients, interruptions and simultaneously working in more than one record. |
| Isolation of patients for infection control increases preventable adverse events.29 | Probably because doctors and nurses directly assess patients less frequently.29 |
| COPD care management trial terminated due to threefold increased all-cause mortality.57 | Unclear mechanism—could have been chance, but increased risk was substantial (threefold) and had a p value of 0.003, well below conventional threshold of 0.05. |
COPD, chronic obstructive pulmonary disease; CPOE, computerised physician order entry; ICU, intensive care unit.