Improvement in Patient Safety May Precede Policy Changes: Trends in Patient Safety Indicators in the United States, 2000–2013

Abstract

Objectives:

Quality and safety improvement are global priorities. In the last two decades, the United States has introduced several payment reforms to improve patient safety. The Agency for Healthcare Research and Quality (AHRQ) developed tools to identify preventable inpatient adverse events using administrative data, patient safety indicators (PSIs). The aim of this study was to assess changes in national patient safety trends that corresponded to U.S. pay-for-performance reforms.

Methods:

This is a retrospective, longitudinal analysis to estimate temporal changes in 13 AHRQ’s PSIs. National inpatient sample from the AHRQ and estimates were weighted to represent a national sample. We analyzed PSI trends, Center for Medicaid and Medicare Services payment policy changes, and Inpatient Prospective Payment System regulations and notices between 2000 and 2013.

Results:

Of the 13 PSIs studied, 10 had an overall decrease in rates and 3 had an increase. Joinpoint analysis showed that 12 of 13 PSIs had decreasing or stable trends in the last 5 years of the study. Central-line blood stream infections had the greatest annual decrease (−31.1 annual percent change between 2006 and 2013), whereas postoperative respiratory failure had the smallest decrease (−3.5 annual percent change between 2005 and 2013). With the exception of postoperative hip fracture, significant decreases in trends preceded federal payment reform initiatives.

Conclusions:

National in-hospital patient safety has significantly improved between 2000 and 2015, as measured by PSIs. In this study, improvements in PSI trends often proceeded policies targeting patient safety events, suggesting that intense public discourses targeting patient safety may drive national policy reforms and that these improved trends may be sustained by the Center for Medicare and Medicaid Services policies that followed.

The Institute of Medicine’s report To Err Is Human: Building a Safer Health System in 1999 highlighted the importance of examining quality in medicine and ushered in a new era of healthcare quality research.¹ Since then, tools have been developed to assess quality of care and identify areas for quality improvement. One such tool is the Agency for Healthcare Research and Quality’s (AHRQ) Patient Safety Indicators (PSIs),² which use administrative data to identify potential adverse inpatient events that may be preventable. Increased PSI rates are often associated with increased mortality, cost, length of stay, and readmission.^3–5 Increasing attention to the human and fiscal cost of low-quality healthcare has led to several changes in federal payment policy related to PSIs.

Since the Institute of Medicine’s report, numerous national and local initiatives targeting patient safety were carried out, particularly numerous reforms from the Center for Medicare and Medicaid Services (CMS). However, earlier reports demonstrate these initiatives have had mixed results. For instance, the Deficit Reduction Act of 2005 authorized the Secretary of Health and Human Services to identify conditions that were reasonably preventable, high-cost or high-volume, and identifiable through International Classification of Disease-9 codes; those conditions would not qualify patients for a higher-severity diagnostic related group (DRG), effectively making the complications nonreimbursable. Many of these identified hospital-acquired conditions (HACs) were previously defined by the National Quality Forum as serious reportable adverse events.⁶ Moreover, enhanced initiatives, such as the Reporting Hospital Quality Data for Annual Payment Update (RHQDAPU) program, now known as the Hospital Inpatient Quality Reporting Program (IQR), include a financial penalty for hospitals not reporting their quality outcomes (CMS 2006).⁷

Another critical step for patient safety improvement came from the Patient Protection and Affordable Care Act (ACA) of 2010, which included a number of pay-for-performance reforms,⁸ including the Hospital Value-Based Purchasing Program, which provided bonuses to hospitals providing high-quality care as determined by their IQR quality reports.⁹ The HAC Reduction Program was also established, using AHRQ’s composite PSI #90 to financially penalize the lowest-performing quartile of hospitals nationally.⁹ Individual PSIs have been identified at different times as HACs or used by the IQR and PSI #90 formulae to assess healthcare quality. In that way, PSIs have proven central to many payment reforms since the publication of To Err Is Human aiming to improve health quality.

For both clinicians and policymakers focusing on the future, it is important to know whether patient safety is truly increasing under the quality improvement programs and funding priorities we have been operating with since the 2000s. Previous study has shown declining rates of inpatient adverse events among some conditions, whereas others—particularly those involving surgical intervention—have been more difficult to change.^10,11

Under priorities of recent national healthcare reform, substantial resources have been allocated to improving the quality and safety of healthcare, which has resulted in a myriad of patient safety initiatives implemented nationally and locally—far too many to assess individually at a population level. However, national improvements in key patient safety trends during this period is unknown are unknown. The aims of this study were to use a nationally representative sample of inpatient discharges to measure changes in national trends of patient safety between 2000 and 2013 and determine whether the changes coincided with major patient safety–related pay-for-performance reforms, which were mostly concentrated between 2008 and 2009. We hypothesized that PSI trends would improve following national patient safety improvement efforts and that there would be an expected lag time between the reforms and changes in hospital practice and eventually in PSI trends. It is important to note that PSIs are a tool for screening and hypotheses generation. Therefore, this study does not attempt to establish a causal relationship between any individual reform and patient safety trends, but rather, the results of this study can inform policy by providing evidence of the extent of impact for value-based payment policies.

METHODS

Data Source and PSIs

We used data from the National Inpatient Sample and Nationwide Inpatient Sample (NIS), Healthcare Cost and Utilization Project, AHRQ, for 2000 to 2013.^12,13 The Nationwide Inpatient Sample, distributed by the Healthcare Cost and Utilization Project, includes up to 15 diagnosis and procedure codes until 2009 and thereafter includes 25 diagnoses. It contains data from 7 million annual hospitalizations in 44 states, a stratified sample of approximately 20% of all inpatient stays in nonfederal hospitals.

The PSIs were developed by AHRQ and are derived from administrative hospital data through standardized algorithms that flag cases with potentially preventable inpatient adverse events attributable to hospital care. The numerator is the number of adverse events of interest and the denominator is the population at risk.² Patient safety indicators have well defined inclusion and exclusion criteria, which helps provide meaning cohort identification. Patient safety indicators were identified using AHRQ PSI software 6.0.1 (September 2016) was used to analyze the PSIs for the years in the study using Statistical Analysis Systems (SAS) Software, Version 9.4.¹⁴ The software was used for the calculation of the PSI risk adjustment coefficients, which are defined as the percentages of discharges that contained a certain PSI out of all discharges that were at risk for that PSI. Certain variables were recoded into uniform formats across all years according to data input requirements of the AHRQ Quality Indicator Software.

Announcement and implementation dates of CMS payment policy changes were identified by searching the CMS Newsroom Web site and Inpatient Prospective Payment Systems Regulations and Notices Web page for final rules containing the string names of selected PSIs.¹⁵

Present on Admission Flag

Since 2007, the CMS has required hospitals to identify a diagnosis that is present on admission (POA) for an inpatient acute care hospital claim.¹⁶ Present on admission codes were implemented into the PSI software in 2009 and have been shown to change PSI rates because they eliminate many of the false positives because of POA. Because our data are from a period antecedent and subsequent to 2009, our analysis did not take into account when a diagnosis was flagged as POA.

Analysis

A total of 13 hospital-level PSIs were obtained from the AHRQ PSI software for each year and are presented in Table 1. Four AHRQ PSIs associated with maternal adverse events were not included in this report (#17, birth trauma; #18, obstetric trauma – vaginal delivery without instrument; #19, obstetric trauma – vaginal delivery with instrument; and #20, Obstetric trauma – cesarean delivery). Three PSIs were not included because of low sample size (#1, complications of anesthesia; #5, foreign body left during procedure; #16, transfusion reaction).

TABLE 1.

Overall Absolute Change and Percent Change in Patient Safety Events, 2000–2013

PSIs	2000	2013	Overall Rate Change	Overall Percent Change
2 – Death in Low-Mortality DRG	0.79	0.41	−0.38	−48.10
3 – Pressure Ulcer	1.44	0.30	−1.14	−79.17
4 – Death Among Surgical Inpatients with Serious Treatable Conditions (failure to rescue)	161.54	117.72	−43.82	−27.13
6 – Iatrogenic Pneumothorax	0.33	0.36	0.03	9.09
7 – Central Venous Catheter–Related Blood Stream Infection	1.85	0.15	−1.70	−91.89
8 – Postoperative Hip Fracture	0.07	0.05	−0.02	−28.57
9 – Perioperative Hemorrhage or Hematoma	4.98	4.97	−0.01	−0.20
10 – Postoperative Physiologic or Metabolic Derangement	0.54	0.63	0.09	16.67
11 – Postoperative Respiratory Failure	11.18	9.80	−1.38	−12.34
12 – Perioperative Pulmonary Embolism or Deep Vein Thrombosis	7.33	4.34	−2.99	−40.79
13 – Postoperative Sepsis	6.66	9.60	2.94	44.14
14 – Postoperative Wound Dehiscence	3.16	1.64	−1.52	−48.10
15 – Accidental Puncture or Laceration	3.04	2.09	−0.95	−31.25

Rates Are per 1000 Discharges.

The AHRQ PSI software uses discharges with International Classification of Diseases, Ninth Revision, Clinical Modification codes corresponding to each PSI and calculates crude, estimated, and risk-adjusted incidence rates. The rates were adjusted for age, sex, DRG, hospital characteristics, and comorbidities based on the Elixhauser index for each PSI.¹⁷ Elixhauser index uses a set of 30 comorbidity measures to distinguish comorbidities with pre-existing diagnosis before admission from a complication acquired during the hospital stay.¹⁷

We evaluated temporal trends in incidence rates using the Joinpoint program (National Cancer Institute, Version 4.5.0.1, https://surveillance.cancer.gov/joinpoint/).

Joinpoint software uses a linear regression approach to study rates over time and determine whether the rates are best described by a single line or multiple linear segments. We allowed a maximum of one joinpoint for each model considered. Joinpoint regression was used to obtain the annual percentage change (APC) of the PSIs over the time.¹⁸ Trends were considered “increasing” or “decreasing” if significantly different from zero (P < 0.05).

RESULTS

Trends in PSI Development

Table 1 shows the overall absolute change and percent change in patient safety events from 2000 to 2013. Initial PSI rates ranged between 0.07 and 161.54 per 1000 discharges. Seven of 13 PSIs had initial rates of greater than 3.0 per 1,000 discharges (range = 3.04–161.54) and 6 PSIs have initial rates ranging between 0.07 and 1.85 per 1,000 discharges. Overall change ranged between −91.9% for central venous catheter–related blood stream infection and 44.1% for postoperative sepsis. Of the 13 PSIs studied, 10 had an overall decrease in rates and 3 had an increase. Among decreasing rates, #2 Death in Low-Mortality DRG, #3 Pressure Ulcer, #7 Central Venous Catheter–Related Blood Stream Infection, #12 Perioperative Pulmonary Embolism or Deep Vein Thrombosis, and #14 Postoperative Wound Dehiscence had an overall decrease of more than 40% (range = − 40.8 to −91.9%), whereas #4 Death Among Surgical Inpatients with Serious Treatable Conditions, #8 Postoperative Hip Fracture, #9 Perioperative Hemorrhage or Hematoma, #11 Postoperative Respiratory Failure, and #15 Accidental Puncture or Laceration showed an overall lower decrease (range = 0% to 31.3%). Three PSIs, #6 Iatrogenic Pneumothorax, #10 Postoperative Physiologic or Metabolic Derangement, and #13 Postoperative Sepsis, showed an overall increase over the study period (9.1–44.1%).

Table 2 reports the APCs of the PSIs studied from 2000 to 2013 with the piecewise joinpoint analysis. Of the 13 PSIs studied, all but one had a significant breakpoint after joinpoint: only #2 Death in Low-Mortality DRG showed a continually steady and significant decrease over the entire study period (APC −5.0). Seven PSIs showed a significant increase followed by a significant decrease in incidence: #3 Pressure Ulcer (APC 2.0 to −29.2), #6 Iatrogenic Pneumothorax (APC 23.5 to −3.0), #9 Perioperative Hemorrhage or Hematoma (APC 1.9 to −4.2), PSI 11 Postoperative Respiratory Failure (APC 3.1 to −3.5), #12 Perioperative Pulmonary Embolism or Deep Vein Thrombosis (APC 8.0 to −13.1); #13 Postoperative Sepsis (APC 7.4 to −6.7); and #15 Accidental Puncture or Laceration (APC 1.6 to −7.8). Two PSIs (#7 Central Venous Catheter–Related Blood Stream Infection and #14 Postoperative Wound Dehiscence) had a steady, even trend followed by a decrease after the breakpoint. Two PSIs, #4 Death Among Surgical Inpatients with Serious Treatable Conditions and #8 Postoperative Hip Fracture, were previously significantly decreasing but have now leveled off in incidence after the breakpoint. Finally, PSI #10 Postoperative Physiologic or Metabolic Derangement showed a significant increase until 2010 and had no changes afterward.

TABLE 2.

Rates of Patient Safety Events as Measured by PSIs and APCs From 2000 to 2013, With Trend Breakpoints

PSIs	Start	End	APC	P
2 – Death in low-mortality DRG	2000	2013	−4.9	<0.001
3 – Pressure Ulcer	2000	2008	2.1	<0.001
	2008	2013	−28.9	<0.001
4 – Death Among Surgical Inpatients with Serious Treatable Conditions (failure to rescue)	2000	2006	−4.4	<0.001
	2006	2013	−0.7	0.097
6 – Iatrogenic Pneumothorax	2000	2002	23.5	0.033
	2002	2013	−3.1	<0.001
7 – Central Venous Catheter–Related Blood Stream Infection	2000	2006	1.3	0.378
	2006	2013	−31.1	<0.001
8 – Postoperative Hip Fracture	2000	2011	−6.1	<0.001
	2011	2013	23.2	0.298
9 – Perioperative Hemorrhage or Hematoma	2000	2009	1.9	<0.001
	2009	2013	−4.2	0.011
10 – Postoperative Physiologic or Metabolic Derangement	2000	2010	3.2	0.005
	2010	2013	−5.8	0.218
11 – Postoperative Respiratory Failure	2000	2005	3.0	0.044
	2005	2013	−3.5	<0.001
12 – Perioperative Pulmonary Embolism or Deep Vein Thrombosis	2000	2006	7.9	0.006
	2006	2013	−13.0	<0.001
13 – Postoperative Sepsis	2000	2009	7.4	<0.001
	2009	2013	−6.8	0.006
14 – Postoperative Wound Dehiscence	2000	2006	−0.9	0.527
	2006	2013	−8.3	<0.001
15 – Accidental Puncture or Laceration	2000	2007	1.6	0.002
	2007	2013	−7.8	<0.001

Changes in PSI-Related Policies

Table 3 shows the major CMS payment policy changes, which occurred during the study period and the joinpoint for each PSI (when present). Three PSIs had no CMS pay-for-performance guidelines developed in the study period: #2 Death In Low-Mortality DRG, #9 Perioperative Hemorrhage or Hematoma, and #10 Postoperative Physiologic or Metabolic Derangement. Four PSIs were classified as HACs: #3 Pressure Ulcer, #7 Central Venous Catheter–Related Blood Stream Infection; PSI #8, Postoperative Hip Fracture; and a special case of PSI #12, Perioperative Pulmonary Embolism or Deep Vein Thrombosis, for select orthopedic procedures. A number of PSIs were integrated in other pay-for-performance programs including the RHQDAPU (now known as IQR) Program, including through inclusion in PSI #90: Patient Safety for Selected Indicators. Except for one PSI (PSI #3, Pressure Ulcer), the announcement or the implementation of payment policy reforms did not precede the declining trends.

TABLE 3.

Major CMS Payment Policy Changes for Selected PSIs With Associated Effective Dates and Joinpoint Years

PSI	Trend Change Year	Trend Direction After Joinpoint	Policy Change Year, Effective Beginning October 1	Policy Change
2 – Death in low-mortality DRG	NA	NA	NA	No relevant policy changes
3 – Pressure Ulcer	2008	Decreasing	2008	Stage 111, IV, and unstageable pressure ulcers deemed to be HACs.
			2009	PSI 3 reporting to IQR required to avoid penalty
4 – Death Among Surgical Inpatients with Serious Treatable Conditions (failure to rescue)	2006	Stable	2009	PSI 4 reporting to IQR required to avoid penalty
6 – Iatrogenic Pneumothorax	2002	Decreasing	2009	PSI 6 reporting to IQR required to avoid penalty
7 – Central Venous Catheter–Related Blood Stream Infection	2006	Decreasing	2008	All vascular catheter-associated infections deemed to be HACs.
			2009	PSI 7 reporting to IQR required to avoid penalty
8 – Postoperative Hip Fracture	2011	Stable	2008	All fractures acquired as an inpatient deemed to be HACs.
			2009	PSI 8 reporting to IQR required to avoid penalty
9 – Perioperative Hemorrhage or Hematoma	2009	Decreasing	NA	No relevant policy changes
10 – Postoperative Physiologic or Metabolic Derangement	2010	Stable	NA	No relevant policy changes
11 – Postoperative Respiratory Failure	2005	Decreasing	2009	PSI 11 reporting to IQR required to avoid penalty
12 – Perioperative Pulmonary Embolism or Deep Vein Thrombosis	2006	Decreasing	2008	Surgical VTE prophylaxis reporting to IQR required to avoid penalty
			2009	pulmonary embolism or deep vein thrombosis after most knee or hip replacement deemed to be HACs
			2009	PSI 12 reporting to IQR required to avoid penalty
13 – Postoperative Sepsis	2009	Decreasing	2009	PSI 13 reporting to IQR required to avoid penalty
14 – Postoperative Wound Dehiscence	2006	Decreasing	2009	PSI 14 reporting to IQR required to avoid penalty
15 – Accidental Puncture or Laceration	2007	Decreasing	2009	PSI 15 reporting to IQR required to avoid penalty

Temporal trends are shown in Figures 1 to 3, according to the policy changes that occurred during the study period. Figure 1 shows PSI trends for PSIs regulated by RHQDAPU/IQR policy reforms (PSI #4, #6, #11, #13, #14, and #15). Figure 2 shows PSI trends for PSIs regulated by both RHQDAPU/IQR and HAC policy reforms (PSI #3, #7, #8, and #12). Figure 3 shows PSI trends for PSIs not regulated by major policy reforms (PSI #2, #9, and #10).

FIGURE 1.

Patient safety indicator trends per 1000 discharges for PSIs regulated by RHQDAPU/IQR policy reform, between 2000 and 2013.

FIGURE 3.

Patient safety indicator trends per 1000 discharges for PSIs not regulated by major policy reform, between 2000 and 2013.

FIGURE 2.

Patient safety indicator trends per 1000 discharges for PSIs regulated by RHQDAPU/IQR and HAC policy reform, between 2000 and 2013.

DISCUSSION

In this study, we evaluated the trends in the rates of patient safety events using PSIs from 2000 to 2013 and considered the timing of pay-for-performance reforms tied associated with each safety event. Our findings showed that overall trends were mostly decreasing (10 of 13), resulting in an absolute safety improvement. A further analysis, carried out with a piecewise linear regression (joinpoint software), found significant annual decreasing or stable trends in all patient safety events for at least the last 5 years of the study period, and importantly, these trends mostly preceded federal payment reform initiatives. Furthermore, there were no patient safety events studied that demonstrated an increase in annual rates, which may reflect worthwhile investment and an improving culture of patient safety and quality in the United States. The average annual percent changes in safety events was ranged from −26%, led by Central Venous Catheter–Related Blood Stream Infections (−91.9%) and trailed by Perioperative Hemorrhage or Hematoma (−0.2%). Safety events with the highest rates and smallest decrease are those related to surgery (e.g., #4 Failure to Rescue, #9 Perioperative Hemorrhage or Hematoma, and #11 Postoperative Respiratory Failure) and therefore represent the greatest opportunity for future improvement.

Given the tremendous amount of resources that have been invested in decreasing adverse patient safety events at both the national and hospital level,¹⁹ this evidence showing consistent trends toward a decline in most events is notable.²⁰ A previous study of PSI data from 1998–2007 using similar methods found an equal number of PSIs increasing as those decreasing in incidence.¹⁰ Indeed, our longer-term trends of decreasing PSI incidence from 2000 to 2013, when viewed in the context of concomitant increased national attention toward adverse events and unnecessary inpatient morbidity, indicate that consistent academic and policymaking focusing on adverse events can produce improvements in patient safety and save lives, consistent with other studies.^21–24

The greatest rate decreases were seen in Pressure Ulcers (PSI #3), Central Catheter-Related Blood Stream Infections (PSI #7), and Perioperative Pulmonary Embolism or Deep Vein Thrombosis (PSI #12). The changes in Pressure Ulcers are hard to interpret, because they coincide with a coding change that excluded low-grade ulcers after 2008.²⁵ However, the changes in Central Catheter–Related Blood Stream Infections and Pulmonary Embolism or Deep Vein Thrombosis rates correspond to substantial efforts in these domains and significant clinical improvement efforts to reduce central line infections and venous thromboembolisms in the past several years.^24,26,27 This is no small matter: interventions required to decrease PSIs, such as central line–associated bloodstream infections (CLABSI), or venous thromboembolism (VTE), for example, amount to a major culture change in both medical and nursing practice. Increased vigilance in central line placement/maintenance or in chemical VTE prophylaxis represent a change in the standard of care in America—one which has successfully improved patient outcomes.^28–30 These efforts are clearly aligned with the decrease in trends we found for PSIs #7 Central Catheter-Related Blood Stream Infections and #12 Deep Vein Thrombosis in particular.

With regard to surgery-related PSIs, such as postoperative respiratory failure and perioperative hemorrhage, improvement programs targeting specific populations have been successfully implemented and specific guidelines have been released.^31,32 However, our findings indicate that surgery–related PSI rates are still high with an overall low percent decrease, suggesting that more effective and systematic safety improvement efforts are needed.

Patient safety indicator–related pay-for-performance policies fall into two categories: those that directly impact payment, such as the HAC policy, and those that improve transparency through public reporting, such as the RHQDAPU (now the IQR) program. In the latter case, PSI development does not directly impact reimbursement but can drive consumer/patient behavior through negative or positive grading on the Medicare Hospital Compare Web site. Some PSIs are reported directly to IQR, and some are reported as a component of PSI 90: Patient Safety for Selected Indicators. It is also important to note that PSIs are not themselves HACs, but certain PSIs—such as PSI #3 (Pressure Ulcer) or PSI #7 (Catheter-Related Blood Stream Infections)—strongly overlap with HACs.

Pay-for-performance reforms are likely implemented both to drive higher quality clinical practice and to assert that a payer is not willing to reimburse for complications that are likely avoidable. For instance, by definition, to be defined as a HAC, one must demonstrate an event to be a reasonably preventable complication. Therefore, we would expect decreases in the incidence of a PSI before it is deemed a HAC, as health systems implement quality improvement processes and demonstrate that an adverse event such as a PSI is indeed preventable. Our study suggests that changes in payment policy may accelerate PSI rate decreases long term. Other studies suggest that hospitals with a high volume of Medicare patients are more responsive than others to CMS pay-for-performance policies, such as the HACs.³³ It is also possible that there is indeed an accelerated PSI rate decrease after the implementation of payment policy changes but that it is masked by simultaneously increased vigilance in surveillance and reporting that pay-for-performance requires.

Further efforts to drive down PSI events may be directed primarily at surgical care. Almost all studied high-incidence PSIs are surgical in nature. Furthermore, PSIs that do not show a declining trend are also surgical—PSI #4 Death Among Surgical Inpatients with Serious Treatable Conditions, PSI #8 Postoperative Hip Fracture, and PSI #10 Postoperative Physiologic or Metabolic Derangement. Although the latter two are fairly low incidence, there are still a substantial number of surgical patients who are dying with what are estimated to be treatable.^34,35 Given costly investments in new surgical technologies and continued evolution of the surgical safety literature, rates of PSI #4 Death Among Surgical Inpatients with Serious Treatable Conditions in particular should be viewed with great attention by clinicians, policymakers, and patients alike.

For the last two decades, quality measures derived from administrative data, and particularly PSIs, have shown to be a powerful instrument for quality and safety improvement in healthcare in the United States and abroad.^36,37 Evidence from outside of the United States highlighted positive changes in clinical practice and outcomes after their introduction.³⁸ International application of PSIs is a challenge involving several countries and international organizations.^39–42 The Organization for Economic Co-Operation and Development (OECD) has promoted initiatives aimed to collect and compare patient safety events from member countries, including 12 AHRQ’s PSIs in its set of indicators.^43–45

Our results should be interpreted keeping in mind important limitations. First, our analysis is based on administrative data that can be affected by variability and inconsistency in the coding of diagnoses and procedures. However, we used a national database that includes approximately a 20% stratified sample of U.S. hospitals, which is a reliable representation of different U.S. care environments. Second, PSIs may contain a significant number of false positives because of a general bias toward events that have established codes against those that do not have codes. Patient safety indicators should be considered a tool for hypotheses generation and screening and should not be considered as exact measures of national performance on patient safety. However, we believe that PSIs can be considered an important first step in identifying clinical targets for more detailed clinical data exploration. Third, our analysis did not take into account POA information because data were collected before and after its implementation. However, trends in our study showed a change after POA implementation only in 3 of 13 cases (hemorrhage/hematoma, metabolic derangement, and sepsis). Fourth, our study design does not allow us to assess a direct causal connection between change in PSI trends and implementation of PSI-related policy reforms. Other factors outside the scope of this study may have influenced both the PSI development trends and the adoption of payment policy changes, including political considerations and discussions between CMS and stakeholder groups. In conclusion, overall rates of PSIs have generally decreased since 2000, indicating that increased attention to these potentially preventable complications is having measurable impact. The most profound decreases were seen in central venous catheter–related blood stream infections and perioperative pulmonary embolism or deep vein thrombosis: two outcomes that have inspired substantial changes in inpatient practice to avoid. The greatest opportunities for further reductions in PSIs are in postsurgical care, where rates remain high and decreases of the study period were minimal. Interestingly, annual rates of patient safety events begin to decrease before the announcement or inception of related payment reforms. This suggests that factors driving significant declines in patient safety events are not related to payment penalties but rather an overall improvement in the quality of healthcare delivery. Research on PSIs needs to be further pursued and expanded both in the United States and internationally, to provide stakeholders evidence to make healthcare safer.

CONCLUSIONS

This study provides key insights on the U.S. context, where a long-term experience in PSIs development and application can draw a path for international use. Patient safety events showed significant annual decreasing or stable trends for at least the last 5 years of the study period, mostly preceded federal payment reform initiatives. No patient safety events demonstrated an increase in annual rates. Understanding the impact of coding, policies, and incentives on the indicators and ultimately on patient safety is needed to perform proper comparisons at local, national and international level.^39,46