Advertisement

Complication Rates on Weekends and Weekdays in US Hospitals

      Abstract

      Purpose

      Recent studies and anecdotal evidence suggest that patient safety may be compromised on weekends. Our objective was to determine whether rates of complications in hospitals are higher on weekends than on weekdays.

      Methods

      We examined records from 4,967,114 admissions to acute care hospitals in 3 states and analyzed complication rates using the Patient Safety Indicators. We selected 8 indicators that could be assigned to a single day: complications of anesthesia, retained foreign bodies, postoperative hemorrhage, accidental cuts and lacerations during procedures, birth trauma, obstetric trauma during vaginal deliveries with and without instrumentation, and obstetric trauma during cesarean delivery. Odds ratios (ORs) comparing weekends versus weekdays were adjusted for demographics, type of admission, and admission route. In a subgroup analysis of surgical complications, we restricted the population to patients who underwent cardiac or vascular procedures.

      Results

      Four of the 8 complications occurred more frequently on weekends: postoperative hemorrhage (OR 1.07, 95% confidence interval [CI], 1.01-1.14), newborn trauma (OR 1.06, 95% CI, 1.03-1.10), vaginal deliveries without instrumentation (OR 1.03, 95% CI, 1.02-1.04), and obstetric trauma during cesarean sections (OR 1.36, 95% CI, 1.29-1.44). Complications related to anesthesia occurred less frequently on weekends (OR 0.86). Among patients undergoing vascular procedures, surgical complications occurred more frequently on weekends (OR 1.46, 95% CI, 1.16-1.85).

      Conclusions

      Rates of complications are marginally higher on weekends than on weekdays for some surgical and newborn complications, but more significantly for obstetric trauma and for surgical complications involving patients undergoing vascular procedures. Hospitals should work toward increasing the robustness of safeguards on weekends.

      Keywords

      It is commonly suspected that weekends and holidays are dangerous times to get sick. During weekends, a higher proportion of patients are admitted to hospitals through the emergency department,
      • Bell C.M.
      • Redelmeier D.A.
      Mortality among patients admitted to hospitals on weekends as compared with weekdays.
      fewer patients are discharged,
      • Varnava A.M.
      • Sedgwick J.E.
      • Deaner A.
      • Ranjadayalan K.
      • Timmis A.D.
      Restricted weekend service inappropriately delays discharge after acute myocardial infarction.
      and many hospital services are unavailable.
      • Bell C.M.
      • Redelmeier D.A.
      Waiting for urgent procedures on the weekend among emergently hospitalized patients.
      However, the incidence of many medical problems and the need for medical care in hospitalized patients have no preference for the day of the week, and although hospitals attempt to maintain a capacity for preserving life and handling emergencies over weekends, the robustness of hospital safety systems and the redundancy of safeguards may be challenged.
      • Rates of complications are higher on weekends than on weekdays for some surgical and newborn complications.
      • The main surgical complication that occurred more frequently on weekends was postoperative hemorrhage.
      • There was a significant increase in the weekend complications (odds ratio 1.46) in patients undergoing vascular procedures.
      • Obstetric trauma with cesarean sections occurred 36% more frequently on weekends (P <.01).
      Prior studies have measured differences in outcomes of hospital care between weekdays and weekends, with mixed results. Barnett and colleagues
      • Barnett M.J.
      • Kaboli P.J.
      • Sirio C.A.
      • Rosenthal G.E.
      Day of the week of intensive care admission and patient outcomes: a multisite regional evaluation.
      and Ensminger and colleagues
      • Ensminger S.A.
      • Morales I.J.
      • Peters S.G.
      • et al.
      The hospital mortality of patients admitted to the ICU on weekends.
      found modestly increased rates of mortality in patients admitted to the intensive care unit on weekends, whereas Bell and Redelmeier
      • Bell C.M.
      • Redelmeier D.A.
      Mortality among patients admitted to hospitals on weekends as compared with weekdays.
      and Cram and colleagues
      • Cram P.
      • Hillis S.L.
      • Barnett M.
      • Rosenthal G.E.
      Effects of weekend admission and hospital teaching status on in-hospital mortality.
      found an increase in mortality of patients admitted through the emergency departments on weekends for conditions that require immediate care. Staffing levels and working conditions, commonly thought to influence patient safety, also vary between weekends and weekdays. Fewer nurses, senior physicians, and a higher average patient acuity all tax safety mechanisms over weekends and during holidays, and several studies suggest a relationship between staffing and safety.
      • Needleman J.
      • Buerhaus P.
      Nurse staffing and patient safety: current knowledge and implications for action.
      Institute of Medicine
      • Pronovost P.J.
      • Angus D.C.
      • Dorman T.
      • Robinson K.A.
      • Dremsizov T.T.
      • Young T.L.
      Physician staffing patterns and clinical outcomes in critically ill patients: a systematic review.
      • Pronovost P.J.
      • Dang D.
      • Dorman T.
      • et al.
      Intensive care unit nurse staffing and the risk for complications after abdominal aortic surgery.
      • Tarnow-Mordi W.O.
      • Hau C.
      • Warden A.
      • Shearer A.J.
      Hospital mortality in relation to staff workload: a 4-year study in an adult intensive-care unit.
      Other published data do not support the presence of a weekend effect. Gould et al
      • Gould J.B.
      • Qin C.
      • Marks A.R.
      • Chavez G.
      Neonatal mortality in weekend vs weekday births.
      showed insignificant differences in rates of neonatal mortality on weekends (although only after adjusting for birth weight). Arias et al
      • Arias Y.
      • Taylor D.S.
      • Marcin J.P.
      Association between evening admissions and higher mortality rates in the pediatric intensive care unit.
      found increased odds of death in children admitted to the intensive care unit during evenings, but no difference between days of the week, and a study looking at measures of quality in a Spanish emergency department found improved care on weekends.
      • Miro O.
      • Sanchez M.
      • Espinosa G.
      • Milla J.
      Quality and effectiveness of an emergency department during weekends.
      Previous studies that examined mortality of patients admitted on weekends are limited by the difficulty of attributing the outcome to a day of the week. However, certain complications, such as those related to surgical procedures, can be linked to specific events and dates. Nevertheless, in the past it has been difficult to study these phenomena because of the absence of validated measures, and because the scarcity of inpatient complications requires large databases to identify sufficient numbers of cases. Our study was conducted using all admissions to inpatient facilities in 3 states over 3 years and used a new tool developed for identifying complication rates. We hypothesized that rates of complications during weekends would be higher, after controlling for differences between weekend and weekday admissions.

      Methods

      Data Sources

      We collected state administrative inpatient data from 1999 to 2001 for New York and Massachusetts, and from 2000 to 2001 for North Carolina. The data were obtained from the Healthcare Utilization Project’s State Inpatient Databases, which are a compilation of data from participating states containing the universe of those states’ nonfederal hospital discharge abstracts.

      Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project State Inpatient Database Webpage. Available at: http://www.ahcpr.gov/data/hcup/hcupsid.htm. Accessed December 1, 2004.

      These databases have been used extensively in health services and outcomes research.

      Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project State Inpatient Database Webpage. Available at: http://www.ahcpr.gov/data/hcup/hcupsid.htm. Accessed December 1, 2004.

      • Schoenman J.A.
      • Sutton J.P.
      • Kintala S.
      • Love D.
      • Maw R.
      • Machlin S.R.
      • Cohen J.W.
      • Thorpe J.M.
      Measuring inpatient care use in the United States: a comparison across five federal data sources.

      Agency for Healthcare Research and Quality. Publications from the Healthcare Cost and Utilization Project Databases. Available at: http://www.ahrq.gov/data/hcup/hcupref.htm. Accessed December 1, 2004.

      They contain standardized data on hospital admissions, including data on diagnoses and procedures that are obtained directly from providers. Although the data quality has limitations, according to some measures they are considered more reliable than household surveys.
      • Machlin S.R.
      • Cohen J.W.
      • Thorpe J.M.
      Measuring inpatient care use in the United States: a comparison across five federal data sources.
      We chose the 3 states for their geographic diversity and the availability of procedure dates in the inpatient database needed for identification of weekend and weekday complications. Every patient admitted to a nonfederal acute care facility during that period was identified, regardless of whether the patient was discharged, transferred to another institution, or deceased. We obtained patient information that included demographics, admission type, admission route, discharge status, International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes for all available diagnoses and procedures, and dates of all procedures, deliveries, cesarean sections, and births. The data were purchased from the Healthcare Cost and Utilization Project, and the study was approved by the institutional review board at Massachusetts General Hospital.

      Identifying Complications on Weekdays and Weekends

      We used the Agency for Healthcare Research and Quality Patient Safety Indicators to identify patients who had complications or were exposed to a risk of complication on weekends and weekdays. The Patient Safety Indicators program, developed by the Agency for Healthcare Research and Quality and Stanford-University of California at San Francisco Evidence-Based Practice Center, classifies more than 20 medical, surgical, newborn, and obstetric complications using administrative data. For each complication, the Patient Safety Indicators algorithm determines a unique population at risk for the complication (denominator) and then identifies those that experienced the complication of interest (numerator).

      Patient Safety Indicators Manual. Rockville, MD: Agency for Healthcare Research and Quality. Available at: http://www.qualityindicators.ahrq.gov/downloads/psi/psi_guide_rev3.pdf. Accessed December 1, 2004.

      Thus, it identifies rates of complications, where each complication has its own numerator and denominator. Each Patient Safety Indicator was selected after a literature search for relevance and was validated and used in numerous peer-reviewed publications.

      Patient Safety Indicators Manual. Rockville, MD: Agency for Healthcare Research and Quality. Available at: http://www.qualityindicators.ahrq.gov/downloads/psi/psi_guide_rev3.pdf. Accessed December 1, 2004.

      • Miller M.R.
      • Elixhauser A.
      • Zhan C.
      • Meyer G.S.
      Patient safety indicators: using administrative data to identify potential patient safety concerns.
      • Romano P.S.
      • Geppert J.J.
      • Davies S.
      • et al.
      A national profile of patient safety in US hospitals.
      Although the Patient Safety Indicators may lack sensitivity, their specificity is high, especially for surgical complications.
      • Miller M.R.
      • Elixhauser A.
      • Zhan C.
      • Meyer G.S.
      Patient safety indicators: using administrative data to identify potential patient safety concerns.
      The Patient Safety Indicators were constructed to indicate trends of complications, to be used as screens and research tools to highlight possible safety problems. It is in this spirit that we applied them to the study of weekend complication rates.
      After consultation with staff physicians and senior coders, we focused on 8 types of complications for which we could determine a singular date of exposure and complication. An example of a Patient Safety Indicator included in our study is birth trauma. The Patient Safety Indicators identify “birth trauma” as documented hypoxic injury, nerve injuries, and several other types of injuries in low-risk births. We posited that the complication occurred on the date of birth, and we could therefore determine whether the complication occurred on a weekend or a weekday. Another complication is “retained foreign bodies.” Here we posited that the foreign body was retained during the principal surgery of the patient’s admission, whose date we could determine. Complications for which we could not identify a singular date, for example, nosocomial infections, were not analyzed. Table 1 lists the 8 complications selected for study, the relevant population at risk, and how the date of complication was determined.
      Table 1Description of Patient Safety Indicators Used in Study
      Complication Group (PSI)Exposed PopulationsRule for Assigning Date of Complication
      1 Complications of anesthesia (S)Inc: surgical patientsDate of principal process or intubation
      Exc: anesthetic poisoning AND active drug dependence, active nondependent abuse of drugs, or self-inflicted injury
      2 Retained foreign body (S)Inc: surgical patientsDate of principal procedure
      3 Postoperative hemorrhage (S)Inc: surgical patientsDate of principal procedure
      Exc: obstetric patients
      4 Accidental laceration during a procedure (S)
      • Inc: surgical patients
      • Exc: obstetric patients
      Date of principal procedure
      5 Birth trauma (N)Inc: liveborn infantsDate of birth
      Exc: infants w/subdural or cerebral hemorrhage, preterm infants, injury to skeleton, osteogenesis imperfecta
      6 Obstetric trauma During vaginal delivery with instrumentation (O)Inc: instrument-assisted vaginal deliveriesDate of delivery
      7 Obstetric trauma during vaginal delivery without instrumentation (O)
      • Inc: vaginal deliveries
      • Exc: instrument-assisted deliveries
      Date of delivery
      8 Obstetric trauma during cesarean delivery (O)Inc: cesarean section deliveriesDate of delivery
      S=surgical complication; N=newborn complication; O=obstetric complication.
      Source: Patient Safety Indicators Ver 2.1 Rev 1.

      Statistical Analysis and Control for Confounding

      Weekends were defined as any Saturday or Sunday (the period from midnight between Friday and Saturday until midnight between Sunday and Monday). In a separate analysis, we included all major federal holidays as “weekends,” but that analysis did not alter the results.
      The data from all states were combined for the analysis to allow sufficient power to detect changes in rates of relatively rare events. We identified populations at risk for each of the 8 types of complication and then calculated rates of complications for each.
      Control for confounding is critical to the attribution of effects when using administrative data. We controlled for the relative heterogeneity in the risk pools with all measurable admission characteristics, as well as comorbidities. We followed the methods established by Cram et al
      • Cram P.
      • Hillis S.L.
      • Barnett M.
      • Rosenthal G.E.
      Effects of weekend admission and hospital teaching status on in-hospital mortality.
      and Bell and Redelmeier,
      • Bell C.M.
      • Redelmeier D.A.
      Mortality among patients admitted to hospitals on weekends as compared with weekdays.
      who found that controlling for route and type of admission minimized the differences between patients admitted on weekends and weekdays. Admission route refers to whether the patient was admitted through the emergency department, and admission type indicates whether the admission was an urgent admission. Comorbidity was measured using the Elixhauser comorbidity method, which accounts for the influence of selected comorbidities on patient outcomes.
      • Elixhauser A.
      • Steiner C.
      • Harris D.R.
      • Coffey R.M.
      Comorbidity measures for use with administrative data.
      • Stukenborg G.J.
      • Wagner D.P.
      • Connors Jr, A.F.
      Comparison of the performance of two comorbidity measures, with and without information from prior hospitalizations.
      Individual comorbidities were used as independent adjusters in the logistic regressions.
      Our principal analytic approach used logistic regression to estimate odds ratios adjusted for demographic characteristics, comorbidities, and admission characteristics (route and type).
      • Rosen A.K.
      • Rivard P.
      • Zhao S.
      • et al.
      Evaluating the patient safety indicators: how well do they perform on Veterans Health Administration data?.
      We used direct standardization to the demographic and comorbidity characteristics of the full study cohort and accounted for clustering of patients within hospitals.
      • Little R.
      Direct standardization: a tool for teaching linear models for unbalanced data.
      • Daniels M.J.
      • Gatsonis C.
      Hierarchical polytomous regression models with applications to health services research.
      To test the stability of our findings, we used propensity scores as an alternative method to control for confounders. This technique analyzes treatment effects in observational studies where randomization is not possible.
      • Daniels M.J.
      • Gatsonis C.
      Hierarchical polytomous regression models with applications to health services research.
      • Rosenbaum P.R.
      • Rubin D.B.
      The central role of the propensity score in observational studies for causal effects.
      • Rubin D.
      Estimating causal effects from large data sets using propensity scores.
      • Rosenbaum P.R.
      • Rubin D.B.
      The bias due to incomplete matching.
      The technique uses measured characteristics to construct a propensity score, which predicts group membership rather than the dependent variable and then generates a control group that matches the case group on key characteristics. We calculated a propensity score using the same adjusters and then used the scores as predictors.
      • Daniels M.J.
      • Gatsonis C.
      Hierarchical polytomous regression models with applications to health services research.
      • Hirano K.
      • Imbens G.
      • Ridder G.
      Because the results of both methods were similar, we only present the numeric results of the logistic regression.
      In a subgroup analysis, we hypothesized that patients undergoing procedures that require specialized surgeons or involve a high level of complexity would be more adversely affected by weekend working conditions. That is, if rates of complications depend on the demands of patient care and the supply of resources and staffing, then the weekend effect might be most pronounced when the demands are high and the supply is short. Patients undergoing vascular and cardiac procedures were selected for analysis because these are sensitive to staffing levels, have a relatively high degree of complexity, and are common. By using ICD-9 codes, we selected the most frequent vascular surgical procedures (ICD-9 codes 38.7 [inferior vena cava filter placement], 39.49 [thrombectomy], 38.12 [carotid endarterectomy], 39.29 [femoropopliteal bypass], 39.27, and 39.42 [new hemodialysis shunt placement]) and most frequent cardiac procedures (ICD-9 codes 36.13, 36.14, 36.15, 36.12 [various coronary artery bypass graftings], 36.01, 36.05, and 36.02 [percutaneous coronary interventions], 37.83 [pacemaker or implantable cardioverter defibrillator device placement], and 37.72 [temporary pacing wires]), and restricted our cohort to those who underwent these procedures. We then performed a logistic regression on these subgroups. To allow for sufficient power, and because the surgical risk pools have similar denominators, we combined the rates for all 4 surgical Patient Safety Indicators. We then calculated differences in rates as the odds ratio of having a complication, with significance at the 95% level.

      Results

      Characteristics of the Population

      We collected data on 4,967,114 admissions of patients at risk for at least 1 of the 8 study complications during our study period. The baseline characteristics of patients who were at risk for a complication on weekends compared with weekdays are shown in Table 2. Patients admitted on weekends constituted 14.8% of the total number of admissions and, on average, were younger and less likely to be white. Table 3 shows the distribution of complications by type and state. We detected 114,090 complications within the population (2.3% of admissions), of which 28.3% were surgical complications, 5.7% were newborn complications, and 66% were related to obstetric trauma. New York was the largest state in the study, with 61% of admissions and 52% of complications.
      Table 2Characteristics of Patients at Risk for Selected Study Complications: New York, North Carolina, and Massachusetts, 1999 to 2001
      WeekendWeekdayTotal
      Surgical admissions
       No.256,0842,678,1182,934,202
       Average age of risk group47.754.253.6
       % Male36.7%41.3%40.8%
       % White60.4%65.1%64.7%
       % Black12.3%10.2%10.4%
       % Hispanic6.6%4.9%5.1%
       Average No. of comorbidities0.450.400.41
       Urgent ED admissions
      Unscheduled admissions through the emergency department.
      24.1%53.5%26.7%
      Newborns
       No.267,395908,1031,175,498
       % Male51.3%51.2%51.3%
       % White48.0%51.6%50.8%
       % Black14.5%13.2%13.5%
       % Hispanic10.2%9.2%9.5%
      Vaginal deliveries
       No.211,342655,865867,207
       Average age of risk group27.728.027.9
       % White48.1%50.9%50.3%
       % Black13.9%13.0%13.3%
       % Hispanic10.0%9.3%9.5%
       Urgent ED admissions
      Unscheduled admissions through the emergency department.
      12.6%11.0%11.4%
       Average No. of comorbidities0.090.090.09
      Cesarean sections
      Admissions with cesarean sections were analyzed for surgical complications and complications specific to cesarean sections.
       No.50,423241,528291,951
       Average age of risk group29.330.029.9
       % White47.6%52.4%51.6%
       % Black16.2%14.0%14.4%
       % Hispanic10.3%9.3%9.5%
       Average No. of comorbidities0.170.150.16
       Urgent ED admissions
      Unscheduled admissions through the emergency department.
      11.1%7.8%8.4%
      Total No. of admissions
      The total number of admissions is less than the sum of the risk pools, because some patients were at multiple risk pools.
      733,3754,233,7394,967,114
      ED=emergency department.
      Source: Healthcare Cost and Utilization Project SID data for New York (1999-2001), Massachusetts (1999-2001), and North Carolina (2000-2001), and Patient Safety Indicators Ver 2.1 Rev 1.
      low asterisk Admissions with cesarean sections were analyzed for surgical complications and complications specific to cesarean sections.
      The total number of admissions is less than the sum of the risk pools, because some patients were at multiple risk pools.
      Unscheduled admissions through the emergency department.
      Table 3Total Number, Percent, and Rate per 100,000 of Selected Complications by Study State and Aggregate
      ComplicationNew YorkNorth CarolinaMassachusettsAll States
      No.PercentRateNo.PercentRateNo.PercentRateNo.PercentRate
      1 Anesthesia8751.5%493841.3%725382.0%8717971.6%61
      2 Foreign bodies4220.7%241780.6%331410.5%237410.6%25
      3 Postoperative hemorrhage32845.6%2109283.2%19912774.9%22954894.8%212
      4 Cuts and lacerations11,87820.1%761655622.8%1406563421.4%101024,06821.1%931
      5 Birth trauma30955.2%42916695.8%78217376.6%72265015.7%553
      6 OB trauma, vaginal delivery with instrumentation644610.9%22,572363612.7%26,935273710.4%24,44612,81911.2%24,072
      7 OB trauma, vaginal delivery without instrumentation32,19754.5%653514,93452.0%986713,66752.0%801760,79853.3%7479
      8 OB trauma, cesarean section9161.5%5044191.5%7815422.1%95918771.6%643
      Total59,113100.0%28,704100.0%26,273100.0%114,090100.0%
      OB=obstetric.

      Surgical Complications

      Adjusted rates of each type of complication, by weekends and weekdays, are displayed in Table 4. Only 1 of the 4 surgical complications, postoperative hemorrhages, occurred more frequently on weekends (227 vs 212 per 100,000). We found no significant differences in the rates of retained foreign bodies or accidental lacerations. Only complications related to administration of anesthesia occurred less frequently on weekends compared with weekdays.
      Table 4Adjusted
      Odds of complication rates were adjusted by logistic regression for age, sex, race, comorbidities, and mode of arrival to the hospital, except complication 5 for sex and race only, and complications 6 to 8 not adjusted for sex.
      Complication Rates per 100,000 Admissions, by Weekend Versus Weekday Occurrence
      ComplicationAdjusted Weekend RateAdjusted Weekday RateOdds Ratio (95% CI)
      1 Anesthesia54630.86
      P <.05.
       (0.78-0.95)
      2 Foreign bodies25260.96 (0.82-1.11)
      3 Postoperative hemorrhage2272121.07
      P <.05.
       (1.01-1.14)
      4 Cuts and lacerations9349470.99 (0.95-1.02)
      5 Birth trauma6005651.06
      P <.05.
       (1.03-1.10)
      6 Vaginal delivery with instrumentation24,35924,3551.00 (0.98-1.02)
      7 Vaginal delivery without instrumentation784076501.03
      P <.05.
       (1.02-1.04)
      8 Cesarean section8526261.36
      P <.05.
       (1.29-1.44)
      Total No. of complications21,48092,610114,090
      CI=confidence interval.
      low asterisk Odds of complication rates were adjusted by logistic regression for age, sex, race, comorbidities, and mode of arrival to the hospital, except complication 5 for sex and race only, and complications 6 to 8 not adjusted for sex.
      P <.05.
      For patients undergoing vascular procedures, we found a significant increase in the weekend rates of complications (OR 1.46). That increase was stable whether we used the logistic regression model or the propensity score model of analysis. There was a trend but no significant difference in the rates of complications in patients with cardiac procedures (OR 1.12).

      Newborn and Obstetric Complications

      Complication rates for birth trauma and 2 of the 3 obstetric trauma indicators were significantly greater on weekends than on weekdays. Complications related to newborn trauma and vaginal deliveries without instrumentation occurred more frequently on weekends (P<.05). The largest effect was observed with obstetric trauma after cesarean sections, 36% more frequent on weekends (P<.01).

      Discussion

      This study aimed to add to the existing debate about the safety of hospital care on weekends by looking at specific complications rather than mortality. We analyzed data from approximately 5 million hospital admissions in 3 states and found small but significantly increased rates of several types of complications on weekends for surgical, newborn, and obstetric patients. We also found complications related to anesthesia occurred less frequently on weekends, and 3 complications for which there were no differences.
      Overall, the rates of adverse events we found were consistent with published literature, although they represent only a subset of the Patient Safety Indicators.
      • Brennan T.A.
      • Leape L.L.
      • Laird N.M.
      • et al.
      Incidence of adverse events and negligence in hospitalized patients Results of the Harvard Medical Practice Study I.
      • Thomas E.J.
      • Burstin H.R.
      • Orav E.J.
      • et al.
      Incidence of and risk factors for adverse events and negligent care in Colorado and Utah in 1992.
      Our results show that, for surgical admissions, there is a small but significantly increased risk of postoperative hemorrhage for operations performed on weekends. Otherwise, there was no overall increased risk of complications on weekends in patients undergoing surgical procedures. However, the risk for patients undergoing vascular procedures on weekends was 46% higher. In addition, we found a greater risk of newborn and obstetric complications on weekends, most notably a 36% increase in risk of complications related to cesarean sections.
      Although we cannot definitively conclude that quality of care is compromised on weekends, several points are worth mentioning. First, we found a meaningful increase in the rates of complications involving vascular surgeries, which is consistent with our hypothesis that surgeries requiring specialized and complex medical care are more sensitive to weekend working conditions. Second, the increase in complication rates of cesarean sections was stable after adjusting for urgent admissions and measured case-mix variables. Although unmeasured variables could still confound the analysis, 1 explanation is that complications among urgent, high-risk cesarean sections are related to staffing and skill levels, which may be compromised on weekends. This is a potentially concerning finding. Third, our modest findings should be interpreted in light of the fact that we examined only 8 types of complications out of dozens of known complications that occur daily in US hospitals. Thus, a small effect may be an underestimate of the true magnitude of the weekend effect as a whole, and improved measurement techniques in the future may enable better quantification. Finally, our finding of reduced rates of anesthesia complications is interesting. It is known that anesthesiologists have been leaders at identifying factors such as production pressure and communication failures, and as a result, they have dramatically decreased the risks of anesthetic death and brain damage during the last 20 years.
      • Cullen D.J.
      • Nemeskal A.R.
      • Cooper J.B.
      • Zaslowky A.
      • Dwyer M.J.
      The effect of pulse oximetry, age, ASA, and clinical status on severity of anesthesia complications: an outcome analysis.
      • Keenan R.L.
      • Boyan C.P.
      Cardiac arrest due to anesthesia A study of incidence and causes.
      • Gaba D.M.
      • Maxwell M.
      • DeAnda A.
      Anesthetic mishaps: breaking the chain of accident evolution.
      • Gaba D.M.
      Human error in anesthetic mishaps.
      Perhaps their processes of care are even more effective on weekends, when surgeries are fewer in number.
      What can be done to improve weekend care? Making US hospitals “7 day-a-week” operations is both costly and unpopular at a time when physicians are searching for higher quality of life and policy makers are searching for ways to curb health care costs. On the other hand, expanding hospital operations may, in fact, be safer and cost-effective. Hospitals are expensive entities with high fixed costs, and weekend “down time” is a waste of resources.
      • Weissman J.S.
      • Bendavid E.
      Should U.S. hospitals go 24/7?.
      Recent investigations have suggested that operating during weekends may enhance patient satisfaction and increase hospital incomes.
      • Eustance C.
      • Carter N.
      • O’Doherty M.
      • Coakley A.J.
      Effect on patient management of a weekend ‘on-call’ nuclear medicine service.
      • Bell C.M.
      • Redelmeier D.A.
      Enhanced weekend service: an affordable means to increased hospital procedure volume.
      Several important limitations of our study are worth mentioning.
      • Gogel H.
      Mortality among patients admitted to hospitals on weekends as compared with weekdays.
      • Halm E.
      • Chassin M.
      Why do hospital death rates vary?.
      There are 2 systematic sources of bias that may apply here: a case-mix bias and a triage effect. The case-mix bias suggests that there are differences in the patient populations between the weekend and weekday cohorts that cannot be detected or controlled using administrative data. It is related to the triage effect, which suggests that hospitals may defer all but the most acute procedures to weekdays, so that patients who are admitted or who undergo operation on the weekend are, in general, more ill than similar patients admitted on weekdays.
      • Dobkin C.
      We believe (supported by previous literature) that by adjusting for case-mix and controlling for route and type of admission, we eliminated most of this bias.
      We used the State Inpatient Databases as the source of information for the Patient Safety Indicators. The State Inpatient Databases are subject to common data quality issues in administrative data sets that rely of physician documentation and coder reliability, including errors in diagnostic coding, missing codes, absence of clinical nuances, and lack of notation for whether the diagnosis was present on admission
      • Miller M.R.
      • Elixhauser A.
      • Zhan C.
      • Meyer G.S.
      Patient safety indicators: using administrative data to identify potential patient safety concerns.
      ; however, there is no known difference between coding of weekend and weekday data that may systematically bias this study. Finally, our date assignment rules introduced imprecision into our analysis, but any bias was canceled out by equal imprecision introduced to the weekend and weekday cohorts.
      Our study leaves several questions for future research: What about other types of complications? How many complications would be prevented by reducing the weekend complication rates? What is the cost of expanding hospital services to 7 days per week? The feasibility and cost-benefit arguments remain open issues.

      Conclusion

      We present evidence that weekend care affects the rates of few complications in acute care hospitals. This increase is mostly small but pronounced for cesarean sections and vascular procedures. We believe it may be explained by hospital staffing structures and resource use. However, although changes to these underlying issues occur slowly, hospitals and some health care providers should be aware of the increased weekend rates of complications and take steps to improve patient safety.

      References

        • Bell C.M.
        • Redelmeier D.A.
        Mortality among patients admitted to hospitals on weekends as compared with weekdays.
        N Engl J Med. 2001; 345: 663-668
        • Varnava A.M.
        • Sedgwick J.E.
        • Deaner A.
        • Ranjadayalan K.
        • Timmis A.D.
        Restricted weekend service inappropriately delays discharge after acute myocardial infarction.
        Heart. 2002; 87: 216-219
        • Bell C.M.
        • Redelmeier D.A.
        Waiting for urgent procedures on the weekend among emergently hospitalized patients.
        Am J Med. 2004; 117: 175-181
        • Barnett M.J.
        • Kaboli P.J.
        • Sirio C.A.
        • Rosenthal G.E.
        Day of the week of intensive care admission and patient outcomes: a multisite regional evaluation.
        Med Care. 2002; 40: 530-539
        • Ensminger S.A.
        • Morales I.J.
        • Peters S.G.
        • et al.
        The hospital mortality of patients admitted to the ICU on weekends.
        Chest. 2004; 126: 1292-1298
        • Cram P.
        • Hillis S.L.
        • Barnett M.
        • Rosenthal G.E.
        Effects of weekend admission and hospital teaching status on in-hospital mortality.
        Am J Med. 2004; 117: 151-157
        • Needleman J.
        • Buerhaus P.
        Nurse staffing and patient safety: current knowledge and implications for action.
        Int J Qual Health Care. 2003; 15: 275-277
        • Institute of Medicine
        Keeping Patients Safe: Transforming the Work Environment of Nurses. National Academies Press, Washington, DC2003
        • Pronovost P.J.
        • Angus D.C.
        • Dorman T.
        • Robinson K.A.
        • Dremsizov T.T.
        • Young T.L.
        Physician staffing patterns and clinical outcomes in critically ill patients: a systematic review.
        JAMA. 2002; 288: 2151-2162
        • Pronovost P.J.
        • Dang D.
        • Dorman T.
        • et al.
        Intensive care unit nurse staffing and the risk for complications after abdominal aortic surgery.
        Eff Clin Pract. 2001; 4: 199-206
        • Tarnow-Mordi W.O.
        • Hau C.
        • Warden A.
        • Shearer A.J.
        Hospital mortality in relation to staff workload: a 4-year study in an adult intensive-care unit.
        Lancet. 2000; 356: 185-189
        • Gould J.B.
        • Qin C.
        • Marks A.R.
        • Chavez G.
        Neonatal mortality in weekend vs weekday births.
        JAMA. 2003; 289: 2958-2962
        • Arias Y.
        • Taylor D.S.
        • Marcin J.P.
        Association between evening admissions and higher mortality rates in the pediatric intensive care unit.
        Pediatrics. 2004; 113: e530-e534
        • Miro O.
        • Sanchez M.
        • Espinosa G.
        • Milla J.
        Quality and effectiveness of an emergency department during weekends.
        Emerg Med J. 2004; 21: 573-574
      1. Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project State Inpatient Database Webpage. Available at: http://www.ahcpr.gov/data/hcup/hcupsid.htm. Accessed December 1, 2004.

        • Schoenman J.A.
        • Sutton J.P.
        • Kintala S.
        • Love D.
        • Maw R.
        The Value of Hospital Discharge Databases: Final Report Under Contract Number 282-98-0024 (Task Order Number 5). NORC at the University of Chicago, Bethesda, MD2005
        • Machlin S.R.
        • Cohen J.W.
        • Thorpe J.M.
        Measuring inpatient care use in the United States: a comparison across five federal data sources.
        J Econ Soc Meas. 2000; 26: 141-151
      2. Agency for Healthcare Research and Quality. Publications from the Healthcare Cost and Utilization Project Databases. Available at: http://www.ahrq.gov/data/hcup/hcupref.htm. Accessed December 1, 2004.

      3. Patient Safety Indicators Manual. Rockville, MD: Agency for Healthcare Research and Quality. Available at: http://www.qualityindicators.ahrq.gov/downloads/psi/psi_guide_rev3.pdf. Accessed December 1, 2004.

        • Miller M.R.
        • Elixhauser A.
        • Zhan C.
        • Meyer G.S.
        Patient safety indicators: using administrative data to identify potential patient safety concerns.
        Health Serv Res. 2001; 36: 110-132
        • Romano P.S.
        • Geppert J.J.
        • Davies S.
        • et al.
        A national profile of patient safety in US hospitals.
        Health Aff. 2003; 22: 154-166
        • Elixhauser A.
        • Steiner C.
        • Harris D.R.
        • Coffey R.M.
        Comorbidity measures for use with administrative data.
        Med Care. 1998; 36: 8-27
        • Stukenborg G.J.
        • Wagner D.P.
        • Connors Jr, A.F.
        Comparison of the performance of two comorbidity measures, with and without information from prior hospitalizations.
        Med Care. 2001; 39: 727-739
        • Rosen A.K.
        • Rivard P.
        • Zhao S.
        • et al.
        Evaluating the patient safety indicators: how well do they perform on Veterans Health Administration data?.
        Med Care. 2005; 43: 873-884
        • Little R.
        Direct standardization: a tool for teaching linear models for unbalanced data.
        Am Stat. 1982; : 38-43
        • Daniels M.J.
        • Gatsonis C.
        Hierarchical polytomous regression models with applications to health services research.
        Stat Med. 1997; 16: 2311-2325
        • Rosenbaum P.R.
        • Rubin D.B.
        The central role of the propensity score in observational studies for causal effects.
        Biometrika. 1983; : 70
        • Rubin D.
        Estimating causal effects from large data sets using propensity scores.
        Ann Intern Med. 1998; 127: 757-763
        • Rosenbaum P.R.
        • Rubin D.B.
        The bias due to incomplete matching.
        Biometrics. 1985; 41: 103-116
        • Hirano K.
        • Imbens G.
        • Ridder G.
        Efficient estimation of average treatment effects using the Estimated Propensity Score. National Bureau of Economic Research Technical Working Paper, NY2000: 251
        • Brennan T.A.
        • Leape L.L.
        • Laird N.M.
        • et al.
        Incidence of adverse events and negligence in hospitalized patients.
        N Engl J Med. 1991; 324 ([see comments.]): 370-376
        • Thomas E.J.
        • Burstin H.R.
        • Orav E.J.
        • et al.
        Incidence of and risk factors for adverse events and negligent care in Colorado and Utah in 1992.
        J Gen Intern Med. 1997; 12: 81
        • Cullen D.J.
        • Nemeskal A.R.
        • Cooper J.B.
        • Zaslowky A.
        • Dwyer M.J.
        The effect of pulse oximetry, age, ASA, and clinical status on severity of anesthesia complications: an outcome analysis.
        Anesthesiology. 1990; 73: A1249
        • Keenan R.L.
        • Boyan C.P.
        Cardiac arrest due to anesthesia.
        JAMA. 1985; 253: 2373-2377
        • Gaba D.M.
        • Maxwell M.
        • DeAnda A.
        Anesthetic mishaps: breaking the chain of accident evolution.
        Anesthesiology. 1987; 66: 670-676
        • Gaba D.M.
        Human error in anesthetic mishaps.
        Int Anesthesiol Clin. 1989; 27: 137-147
        • Weissman J.S.
        • Bendavid E.
        Should U.S. hospitals go 24/7?.
        Am J Med. 2004; 117: 202-203
        • Eustance C.
        • Carter N.
        • O’Doherty M.
        • Coakley A.J.
        Effect on patient management of a weekend ‘on-call’ nuclear medicine service.
        Nucl Med Commun. 1994; 15: 388-391
        • Bell C.M.
        • Redelmeier D.A.
        Enhanced weekend service: an affordable means to increased hospital procedure volume.
        CMAJ. 2005; 172: 503-504
        • Gogel H.
        Mortality among patients admitted to hospitals on weekends as compared with weekdays.
        N Engl J Med. 2002; 346: 1500
        • Halm E.
        • Chassin M.
        Why do hospital death rates vary?.
        N Engl J Med. 2001; 345: 692-694
        • Dobkin C.
        Hospital staffing and inpatient mortality. Department of Economics, University of California at Berkeley, Mimeo2003