Quality and Technology N9205 Oct. 17, 2000 Assessing the quality of care or services Was the right thing done? Was it done done right? Did it yield the right results? Columbia University School of Nursing M6920, Fall, 2000 Donabedian framework Structure/input • capital investment • staffing • relationships Process • content • sequence Outcome Columbia University School of Nursing M6920, Fall, 2000 Assessing quality Person seeks care Provider Primary Preventio n Outreach Activities Case Finding Screening Diagnosis Diagnosis Management Evaluation of Presenting History,Physical Complaint Other Diagnostic Procedures Patient Education Referrals Therapy Monitoring Followup Columbia University School of Nursing Desired effects Office of Technology Assessment, 1988 M6920, Fall, 2000 Critical issues Selection of domain Selection of measures Identification of data source Columbia University School of Nursing M6920, Fall, 2000 A special case : technology assessment Generally includes "machines" Would also cover pharmaceuticals? Other possible "hidden" technologies • scheduling • staffing patterns • access systems Columbia University School of Nursing M6920, Fall, 2000 Use of technologies? • clinical excellence • technological preeminence • profit maximization • in a fee-for-service system • in a capitated or global budget system Columbia University School of Nursing M6920, Fall, 2000 Assessing technology Is this safe? Efficacious? Effective? Efficient? • speed of outcome • quality of outcome • cost of outcome Columbia University School of Nursing M6920, Fall, 2000 Renal dialysis introduction-late 60's/early 70's use of screening committees ESRD Medicare policy US compared to GB Columbia University School of Nursing M6920, Fall, 2000 Heart transplant early 70’s • everybody try one • few centers persist procedure with mid 80's • introduction of anti-rejection drugs Columbia University School of Nursing M6920, Fall, 2000 CABG surgery what are the trade-offs in quality of life? what about skill/competence • limitations on facilities performing in NY state Columbia University School of Nursing M6920, Fall, 2000 BC/BS Technology Assessment Agenda for 1997 Cost Effectiveness Analyses • Cervical Cancer Rescreening Methods • Electron beam computed tomography for CHD Columbia University School of Nursing M6920, Fall, 2000 Clinical Effectiveness Analyses • • • • • fetal febrnectin functional sterotactic radiosurgery genetic testing for colon cancer neurostimulation for tremor non-coronary intravascular ultrasound Columbia University School of Nursing M6920, Fall, 2000 Critical policy problems who is "disinterested observer" to conduct assessment? • use of consensus panels (NIH/RAND models) • one discipline? inclusion of "doers"? • OTA elimination; sizing AHCPR defining "experimental"? appeal to the courts Columbia University down- School of Nursing M6920, Fall, 2000 Critical research questions use/role of public opinion professional opinion and practice • too rapid adoption • delayed adoption financial incentives to use/not use short and long-term outcomes Columbia University School of Nursing M6920, Fall, 2000 Hamilton & HO Objective: understand the relationship between volume and quality Reason: Is it “practice makes perfect” or selective referral patterns? Method: regression analysis of 3 years of data Columbia University School of Nursing M6920, Fall, 2000 Hamilton & Ho, Cont. Result: negative relationship between volume and length of stay But: fluctuations in volume had no effect on LOS or mortality Conclusion: high volume = high quality for reasons other than practice makes perfect Columbia University School of Nursing M6920, Fall, 2000 Meehan et al. PRO study to • assess quality of care for Medicare patients with pneumonia • determine whether process of care performance is associated with lower mortality multi-center retrospective cohort study (14,069 patients; 3555 hospitals in US) Columbia University School of Nursing M6920, Fall, 2000 Mehan et al, cont. Definition of process of care • time from arrival to antibiotic administration • blood culture before initial antibiotics • blood culture within 24 hours of hospital arrival • oxygenation assessment within 24 hours Columbia University School of Nursing M6920, Fall, 2000 Mehan et al, cont. Sample Selection • decision on ICD-9-CM codes • exclusion criteria (primarily clinical confounders such as HIV) Data collection • training of medical records abstractors Columbia University School of Nursing M6920, Fall, 2000 Mehan et al, cont. 1/4 of elderly patients do not receive antibiotics until at least 8 hrs post admission; doing so is associated with 15% lower odds of mortality 1/3 of elderly patients do not have a blood culture drawn within 24 hours; doing so associated with 10% lower odds of mortality Columbia University School of Nursing M6920, Fall, 2000 Mehan et al, cont. high rate of unconfirmed pneumonia diagnoses when clinical criteria were included Intriguing query: did presence of DNR orders limit therapy for some patients? Columbia University School of Nursing M6920, Fall, 2000 Mezey et al Cross sectional telephone survey Sample of 1016 from 1452 calls • • • • over 18 English or Spanish speaking medical or surgical admission no nursing home pre or post stay Instrument? Columbia University School of Nursing M6920, Fall, 2000 Mezey et al Forced choice answers? Findings • Racial, language and economic differences • Level of education most significant Columbia University School of Nursing M6920, Fall, 2000 Zinn et al Objective: identify contextual attributes that influence TQM adoption Data: survey of licensed nursing home administrators, certification files and ARF Columbia University School of Nursing M6920, Fall, 2000 Zinn et al, Variables Variable Definition Source Nsg. Home has adopted TQM survey Perceived competition Admin. Perception TQM survey Herfindal index Nsg home market concentration MMACS Excess capacity Average # empty beds/county MMACS Hospital-based substitutes # hospitals providing LTC ARF Nursing home size # beds in facility MMACS M’care market penetration Proportion of discharges Medicare ARF HMO membership Proportion of residents in HMO ARF Proportion Medicare Proportion of NH residents with Medicare coverage MMACS Per capita income (log) Average per capita income in county ARF Dependent Variabl e TQM Adoption Independent Variables Columbia University School of Nursing M6920, Fall, 2000 Zinn et al, cont. 1: more competitive markets lead to adoption--Partial support 3: facilities in areas with higher M’care discharges more likely to adopt--support 4: facilities in areas with greater HMO penetration are more likely to adopt--significant support Columbia University School of Nursing M6920, Fall, 2000 Zinn et al, cont 2: Larger facilities are more likely to adopt--no support 5: Facilities with grated proportion of M’care recipients in total census are more likely to adopt-no support Columbia University School of Nursing M6920, Fall, 2000 Keeler et al How can a good case mix method be developed? Combination of birth certificate and hospital discharge data Retrospective model building effort Columbia University School of Nursing M6920, Fall, 2000 Keeler et al Factors ruled out • race and management decisions Factors had to have • • • • consistent coding practices unequivocally risk not outcome prevalence consistent with clinician view recorded variable associated with outcomes Columbia University School of Nursing M6920, Fall, 2000 Keeler et al Merged data better than only one source Simple model explains 30% of variance among hospitals Best model explains 37% Is the remainder practitioner choice??? Columbia University School of Nursing M6920, Fall, 2000