Patients classification System 
tools for 
tools for 
tools for 

4 ANECDOTE 4 contact 
4 Logic 
4 deaths 

OUTPUT FILES DESCRIPTION 

Overview 
All indicators are provided for hospital acute somatic care. Stays in geriatrics, rehabilitation units, psychiatrics units (M500, M900, M950, M990) are not taken into account. Global results are provided separately for each hospital and hospital site in Excel workbooks specific to each indicator (readmission.xls, cost.xls, etc.). Detailed results are provided in separate text files, corresponding to the eligible population of specific indicators: 
Eligible_discharges.txt for READMISSIONS (excluding
transferred and dead patients, healthy newborns, foreigners, candidates for
day surgery); To facilitate further analyses, SQLape® tool generate four data files:  Cases.txt, with all
non medical variables included in SQLape_input.txt file 

Notation 
A systematic notation is used for all results, with indices describing observed value (index 1) and expected values (index 0, 0min and 0max giving the 95% confidence intervals). The two letters before indices correspond to abbreviations of each indicator:
For instance, LS_{1} is the observed length of stay and RO_{0min} is the minimal expected rate of potentially avoidable reoperations. For each indicator, the SQLape® tool computes the ratio between observed and expected values (rate ratio, ratio of costs, ratio of length of stay). For instance, R_{AR} = R_{1}/R_{0} Performance is measured by ratios, with good results if ratios are low (observed < expected values) and bad results corresponding to high ratios. A ratio of 1.2 means that observed rate is 20% greater than expected, for instance. Expected values are provided with 95% confidence intervals, assessing if they differ significantly from observed rates: A
observed rate < minimal expected rate (significantly lower than expected) Some additional information is provided in global results (per hospital and site) c number of cases (numerator) e elective population size (denominator) n number of stays provided in SQLape_input.txt file, corresponding to acute somatic care q quality of data, measured specifically for each indicator (see below) s sum d difference p potential reduction (proportion of potential improvement) Some additional information is provided in detailed results (per stay or operation): Dates:
admission, discharge, readmission, reoperation dates for instance (01.01.2200
or 31.12.2999 correspond to unspecified dates) 

Context variables 
To assess length of stays, hospital costs and beds, we compute the variable “analysis” describing the context of hospital stays: A.
Back home, not followed by a potentially avoidable readmission (if not
classified as BF) 

New case definition (Switzerland) 
When cases include several hospital stays (new Swiss definition of the case, introduced in 2012), each stay is newly built with admission and discharge dates. The case identifier is extended with the rank of the stay (for instance: 19222110_1, 19222110_2, 19222110_3 if there are two interruptions). Diagnoses are linked to each stay, but procedures are allocated to corresponding stays knowing their dates. 

GLOBAL RESULTS Synthesis (Synthesis.xls) 
Variables Description For each hospital and site: A Indicator 
Potentially avoidable readmission n Number of stays in the SQLape_input.txt file q Quality of data N Numerator of the indicator (cases, hospital days or Swiss francs) D Denominator of the indicator (eligible population) V_{1} Observed value (rate, length, cost) = N/D V_{0} Expected
value (rate, length, cost) V_{0max} Maximal expected rate R Rate, length or cost ratio (observed/expected values) V Result : 

GLOBAL RESULTS Length of stay (Length.xls) 
Variables Description Variables Description H Hospital identifier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{LS} quality of data for length of stay indicator c_{LS} number
of hospital days (including admission and discharge days, except for stays
with interruptions e_{LS} number of discharges LS_{1} observed length of stay (c_{LS}/e_{LS}) LS_{0} expected
rate LS_{0max} maximal expected rate R_{LS} length ratio (LS_{1}/LS_{0}) V_{LS} Result : 

GLOBAL RESULTS Hospital cost (Cost.xls)

Variables Description H Hospital identifier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{CO} quality of data for cost indicator c_{CO} amount of hospital resources (Swiss francs) e_{CO} number of discharges CO_{1} observed cost (c_{CO}/e_{CO}) CO_{0} expected
cost CO_{0max} maximal expected cost R_{CO} cost ratio (CO_{1}/CO_{0}) CO_{0CH} Swiss reference rate CO_{a} Adjusted reoperation rate (CO_{0CH} * R_{CO}) V_{CO} Result : 

GLOBAL RESULTS Potentially avoidable readmissions (Readmission.xls)

Variables Description H Hospital idenfier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{AR} quality of data for readmission indicator c_{AR} number of potentially avoidable readmissions (internal and external) e_{AR} number of eligible discharges AR_{1} observed rate of potentially avoidable readmissions (=AR_{1i} + AR_{1e}=c_{AR}/e_{AR}) AR_{1i} internal observed rate (in the same hospital readmissions) AR_{1e} external observed rate (in other hospital readmissions) AR_{0} expected
rate (internal and external) AR_{0max} maximal expected rate R_{AR} rate ratio (AR_{1}/AR_{0}) v_{AR} Result : 

GLOBAL RESULTS Potentially avoidable reoperations (Reoperation.xls)

Variables Description H Hospital identifier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{RO} quality of data for reoperation indicator c_{RO} number of potentially avoidable reoperations e_{RO} number of eligible operations RO_{1} observed rate of potentially avoidable reoperations (c_{RO}/e_{RO}) RO_{0} expected
rate RO_{0max} maximal expected rate R_{RO} rate ratio (RO_{1}/RO_{0}) V_{RO} Result : 

GLOBAL RESULTS Premature death (Death.xls)

Variables Description H Hospital identifier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{PD} quality of data for premature death indicator c_{PD} number of premature deaths e_{PD} number of discharges of patients not at the end of their life PD_{1} observed rate of premature deaths (c_{PD}/e_{PD}) PD_{0} expected
rate PD_{0max} maximal expected rate R_{PD} rate ratio (PD_{1}/PD_{0}) V_{PD} Result : 

GLOBAL RESULTS Iatragenic complications (Complication.xls) 
Variables Description H Hospital identifier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{CP} quality of data for complication indicator c_{CP} number of iatrogenic complications e_{CP} number of discharges of patients CP_{1} observed rate of iatrogenic complications (c_{CP}/e_{CP}) CP_{0} expected rate CP_{0min} minimal expected rate CP_{0max} maximal expected rate R_{CP} rate ratio (CP_{1}/CP_{0}) V_{CP} Result : A observed rate < minimal expected rate (good) B minimal expected rate < observed rate < maximal expected rate C observed rate > maximal expected rate (too high) 

GLOBAL RESULTS Day surgery (Daysurgery.xls) 
Variables H Hospital identifier L Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) n number of stays in the SQLape_input.txt file q_{DS} quality of data for day surgery (same criteria than for length of stay) c_{DS} number of candidates for one day surgery e_{DS} number of eligible discharges DS_{1} observed rate of candidates for one day surgery (c_{DS}/e_{DS}) DS_{0} expected
rate DS_{0max} maximal expected rate R_{DS} rate ratio (DS_{1}/DS_{0}) V_{DS} Result : 

GLOBAL RESULTS Beds reduction (Beds.xls) 
Variables Description H Hospital identifier Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) N_{1} observed number of stays N_{0} expected
number of stays (if no more than 15% of candidates for one day surgery LS_{1} observed length of stay LS_{0} expected
length of stay d_{LS} Excess
of length of stay d_{LS} = (LS_{1C} – LS_{0C}) 
(LS_{1}_{Æ} – LS_{0}_{Æ}), p_{N} reduction
potential related to case numbers p_{CP} reduction
potential related to complications p_{LS} reduction
potential related to length of stay BE_{1} observed number of beds (=SLS_{1 }/ (365 *0.85) BE_{0} expected number of beds (=SLS_{1 }(1p_{N}P_{CP}P_{LS})_{ }/ (365 *0.85) D_{BE} potential reduction of the number of beds (B_{1} – B_{0})
General explanations There are three ways to reduce the number of hospital days and therefore hospital beds: to avoid unnecessary hospitalizations, to lower the incidence of complications and to reduce length of stay. Avoiding unnecessary hospitalizations (proportion p_{N}) It is not wise to eliminate all candidates for one day surgery, because some of them are justified, for instance because patients live alone, too far from hospital to come back if a complication occur or for other reasons. We consider here that a lower threshold of 15% is normal, but that supernumerary stays could probably be treated in less than 24 hours. Similarly, we consider that a threshold of 5% of unjustified stays is normal, because some patients need to be monitored to exclude severe morbidities or risk of complications. But, many of such situations might also be managed by ambulatory care and the supernumerary stays might provide hospital days savings. The impact of avoiding unnecessary hospitalizations is estimated by the difference between observed and expected numbers of stays (N_{1}N_{0}), weighted by optimal length of stay (the minimum between observed and expected values, min[LS_{0};LS_{1}]). One day was considered appropriate for candidates for one day surgery. This potential savings (p_{N}) is expressed in proportion of actual number of hospital days (sum of observed hospital days LS_{1}).
Lowering complications rates (proportion p_{CP}) Expected lengths of stay are computed considering all illnesses of patients, inclusive complications. This is justified by the fact that not all complications can be avoided even in the best hospitals. But if there is an excess of complications rates (d_{CP}), there is a corresponding potential of reduction, estimated by the observed excess of length due to complications (d_{LS}). Reducing length of stay (proportion p_{LS}) The potential reduction is computed by the difference of lengths of stay (observed LS_{1} – expected LS_{0}) weighted by the observed number of cases (N_{1}) and expressed in proportion of the total number of hospital days (SLS_{1}). The part of excessive length due to complication (p_{CP}, see above) is not considered here. Estimation of the impact on the number of beds The number of beds is estimated with a standard bed occupancy rate of 85% and a full utilization during the all years (365 days). 

DETAILED RESULTS Length of stay, hospital cost and complications (Analysis.txt) 
Variables Description #Case Hospital stay identifier Analysis A.
Back home, without readmission D. Nursing home placement (Stay before <>3 and <>4, Stay after =2 or 3) E. Day surgery F. Unjustified stay #Hospital Hospital identifier #Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) #Patient Patient identifier #Case Hospital stay identifier DateIndex Date of index operation (each day = another eligible operation) Age Age at admission Gender 1 = male, 2 = female Entry 0
= birth, 1 = transfer (Stay before = 5 or 6), 2 = other Specialty 4 SPECIALTY (.pdf) Mission New variable added for hospital planning purpose (more details will be given soon on the website) Category Main SQLape® category CP_{1} Observed
rate of complications (1 = yes, 0 = no) CP_{0min} Minimal expected rate of complications (idem) CP_{0max} Maximum expected of complications (idem) LS_{1} Observed
length of stay LS_{0min} Minimal expected length of stay (idem) LS_{0max} Maximum expected length of stay (idem) LS_{0_CP} Expected length of stay for observed complications CO_{1} Observed
hospital cost (1 = yes, 0 = no) CO_{0min} Minimal expected hospital cost (idem) CO0_{max} Maximum expected hospital cost (idem) 

DETAILED RESULTS Readmissions (Eligible_discharges.txt) 
Variables Description #Hospital Hospital identifier #Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) #Patient Patient identifier #Case Index stay identifier (index stay = eligible for a possible readmission) Age Age at admission Gender 1 = male, 2 = female Previous Hospitalization during the six months prior to admission date Programmed 1=planned, 0=unplanned (corresponding to “admission mode” = 1 or 3 in SQLape_input.txt file) #GroupAR Clinical groups, used to compute expected rates (labels in Excel table) AdmissionDate Admission date of the index stay (each eligible discharge) DischargeDate Discharge date of the index stay (each eligible discharge) AR_{0} Expected rate AR_{0min} Minimal expected rate (idem) AR_{0max} Maximum expected rate (idem) AR_{1} Observed rate (internal or external, = 1 if potentially avoidable readmission, 0 if not ReadmissionDate Admission date of the readmission stay (31.12.2999 if no readmission) ARDelay Admission date of the readmission – Discharge date of the index stay #Readmission Identifier of the readmission ReadmissionHospital Identifier of the hospital of readmission AR_{1i} Observed internal rate (= 1 if potentially avoidable readmission in the same hospital, 0 if not) 

DETAILED RESULTS Reoperations (Eligible_operation.txt) 
Variables Description #Hospital Hospital identifier #Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) #Patient Patient identifier #Case Hospital stay identifier DateIndex Date of index operation (each day = another eligible operation) Age Age at admission Gender 1 = male, 2 = female Programmed Admission mode = 1 or 3 #GroupRO Clinical groups, used to compute expected rates (labels in Excel table) AdmissionDate Admission date of the index stay (each eligible discharge) DischargeDate Discharge date of the index stay (each eligible discharge) RO_{0} Expected rate RO_{0min} Minimal expected rate (idem) RO_{0max} Maximum expected rate (idem) RO_{1} Observed rate (= 1 if potentially avoidable reoperation,0 if not ReoperationDate Admission date of the reoperation RODelay Reoperation date – previous operation date 

DETAILED RESULTS Premature death (Eligible_death.txt) 
Variables Description #Case Hospital stay identifier #Hospital Hospital identifier #Site Location of the hospital Year Year (most frequent discharge year in the SQLape_input.txt file) Specialty 4 SPECIALTY (.pdf) Complexity 3
> 7 SQLape® categories (complications categories excluded) 1 1 SQLape® category (complications categories excluded) AgeClass Class
of age, expressed in years PD_{0} Expected rate PD_{0min} Minimal expected rate (idem) PD_{0max} Maximum expected rate (idem) 

DETAILED RESULTS Candidates for one day surgery (Eligible_surgery.txt) 
Variables Description EligibleDS Hospital stay identifier #Hospital Hospital identifier #Site Location of the hospital DS_{1} Observed rate (= 1 if candidates for one day surgery, 0 if not DS_{0} Expected rate = 10% of eligible stay with surgery DS_{0min} idem (no statistical model) DS_{0max} idem (no statistical model) 

DETAILED RESULTS Unjustified hospitalizations (Eligible_cases.txt) 
Variables Description #Case Hospital stay identifier #Hospital Hospital identifier #Site Location of the hospital UJ_{1} Observed rate (= 1 if candidates for one day surgery, 0 if not) UJ_{0} Expected rate = 10% of eligible stay with surgery UJ_{0min} idem (no statistical model) UJ_{0max} idem (no statistical model) 

© Yves Eggli and SQLape s.à.r.l., 2014. Last update: 29.04.2016 



