# Patient Harm in Hospitals: Why a Decade of Safety Efforts Hasn't Reduced Medical Errors

In a landmark study published in the *New England Journal of Medicine*, researchers found that patient harm from medical care remained alarmingly common in 10 North Carolina hospitals between 2002 and 2007, with **little evidence of improvement** despite a decade of national patient-safety initiatives. The study reviewed over 2,300 hospital admissions and found 588 cases of harm — approximately **25 harms per 100 admissions** — with no statistically significant reduction in overall harm rates over the six-year period. This research, which validated concerns raised after the Institute of Medicine's groundbreaking 1999 report "To Err Is Human," suggests that substantial work remains to translate known safety interventions into routine hospital practice.

# Patient Harm in Hospitals: Why a Decade of Safety Efforts Hasn't Reduced Medical Errors

## Table of Contents

- Key Points
- Background: The "To Err Is Human" Wake-Up Call
- Why Researchers Chose North Carolina
- Study Methods: How the Research Was Conducted
- Hospital Selection and Record Review
- Ensuring Accuracy: Reliability Checks
- Key Findings: The Types and Severity of Patient Harm
- Preventable Harms: What Could Have Been Avoided
- Changes Over Time: Did Patient Safety Improve?
- Clinical Implications: What This Means for Patients
- Study Limitations: What This Research Couldn't Prove
- Recommendations for Patients and Families
- Frequently Asked Questions
- Source Information

## Key Points

- In a study of 2,341 hospital admissions in North Carolina, 588 harms occurred, about 25 per 100 admissions.
- No statistically significant reduction in harm rates occurred from 2002 to 2007 despite national safety initiatives.
- 63.1% of harms were rated preventable, including permanent harm, life-threatening events, and deaths.
- Most common harms were procedure-related, medication-related, and hospital-acquired infections.
- Patients can ask about medications, fall prevention, infection prevention, and pressure ulcer prevention to reduce risk.

## Background: The "To Err Is Human" Wake-Up Call

In December 1999, the Institute of Medicine (IOM) — now known as the National Academy of Medicine — released a landmark report titled *To Err Is Human* that shocked the medical community and the public alike. The report estimated that medical errors cause **up to 98,000 deaths and more than 1 million injuries each year** in the United States.

That report triggered an unprecedented response. Accreditation bodies, health insurance payers, nonprofit organizations, government agencies, and hospitals themselves launched major initiatives and invested considerable resources to improve patient safety.

Some specific interventions have been proven to reduce errors, including:

- Computerized provider order-entry systems (digital systems for ordering medications and tests)
- Limiting residents' (doctors in training) work shifts to no more than 16 consecutive hours
- Implementing evidence-based "care bundles" — groups of proven practices that improve patient outcomes when performed together

However, many of these interventions were never rigorously evaluated, and even proven ones were not reliably implemented on a large scale. The central question remained unanswered: **Did all these efforts actually make patients safer?** This study was designed to find out.

## Why Researchers Chose North Carolina

The researchers deliberately selected North Carolina as the study site because it was a state *likely* to show improvement, if improvement was happening anywhere. North Carolina had demonstrated an unusually high level of engagement in patient-safety efforts, including:

- A **96% rate of hospital enrollment** in a previous national improvement campaign, compared with an average of **78% in other states**

- Extensive participation in statewide safety training programs
- Active involvement in improvement collaboratives (groups of hospitals working together to share best practices)

By choosing a state that was actively engaged in safety work, the researchers ensured that any absence of improvement could not simply be blamed on a lack of effort.

## Study Methods: How the Research Was Conducted

The study applied a tool called the **Global Trigger Tool for Measuring Adverse Events**, developed by the Institute for Healthcare Improvement (IHI). This tool is designed to help reviewers efficiently and reliably detect harm in medical records by looking for specific "triggers" — clues in a patient's chart that suggest medically induced harm may have occurred.

The trigger tool contains **52 distinct triggers**. For example, the administration of naloxone (a medication used to reverse the effects of an accidental narcotic overdose) would signal a possible medication error. When a reviewer found a trigger, they investigated the chart further to determine whether harm had actually occurred.

The study was retrospective, meaning it looked back at medical records of patients who had already been discharged. It included harms that were present when the patient was admitted (resulting from prior treatment) as well as harms that occurred during the studied hospital stay, to capture the total burden of medical harm.

## Hospital Selection and Record Review

All acute care hospitals in North Carolina listed in the American Hospital Association database were eligible, except those providing exclusively pediatric, rehabilitation, or psychiatric care. Hospitals were grouped by size (small, medium, or large), location (urban or rural), and teaching status (teaching or nonteaching).

Researchers invited **14 hospitals** to achieve their goal of **10 participating hospitals** (a 71% participation rate). In each hospital, **10 randomly selected admissions of at least 24 hours** were reviewed for each quarter from January 2002 through December 2007 — a total of **240 records per hospital**.

A critical design element: records were reviewed in **random order** (not by admission date), and **dates of hospitalization were concealed from reviewers**. This prevented bias — for instance, reviewers unconsciously seeing "improvement" because they knew which records came from earlier versus later years.

Two independent teams reviewed the *same* records:

1.
**Internal reviewers** — hospital-based teams (typically nurses as primary reviewers, with physician backup) who worked in the hospitals where they reviewed charts.
1.
**External reviewers** — teams hired and supervised by an outside research organization (Battelle), who worked elsewhere.

Both teams received identical training through standardized web-based seminars, including didactic sessions, practical review exercises, and debriefing sessions. Each record review was completed in **20 minutes or less**.

In the second stage of review, primary reviewers wrote one-to-two paragraph summaries of all suspected harms. Two independent physician reviewers — who were also unaware of hospitalization dates — made final determinations about whether harm occurred, how severe it was, and whether it was preventable.

## Ensuring Accuracy: Reliability Checks

The researchers tested the reliability of their review process through multiple checks. Reliability is measured using a statistical tool called a **kappa score**, where scores range from 0 (no agreement) to 1 (perfect agreement). Scores above 0.6 are generally considered "substantial" agreement, and above 0.8 "almost perfect" agreement.

Internal review teams performed more reliably, with kappa scores for detecting harm ranging from **0.64 (substantial) to 0.93 (almost perfect)** across seven reliability tests. External reviewers scored lower, with kappa scores ranging from **0.40 (moderate) to 0.72 (substantial)**.

For preventability ratings (deciding whether a harm could have been avoided), kappa scores were **0.83 for internal reviewers** and **0.54 for external reviewers**.

The researchers also had a team of expert reviewers — highly experienced in using the trigger tool — independently evaluate a **10% sample of records** from each hospital. Internal reviewers agreed with these experts about the presence of harm in **81% of reviews** (kappa 0.49), while external reviewers agreed in **75% of reviews** (kappa 0.32). Internal reviewers also had better agreement with experts on severity ratings (kappa 0.53 vs. 0.26). In short, **the internal reviewers produced higher-quality data by every measure** — an important detail when interpreting the results.

## Key Findings: The Types and Severity of Patient Harm

Internal teams completed **2,341 of 2,400 planned record reviews (97.5%)** across the 10 hospitals. They identified a total of **588 harms** during **10,415 patient-days** of observation.

This translates to:

-
**56.5 harms per 1,000 patient-days** (95% confidence interval [CI], 52.0 to 61.2)
-
**25.1 harms per 100 admissions** (95% CI, 23.1 to 27.2)
- These harms occurred in **423 unique patient admissions (18.1%)** — meaning nearly 1 in 5 admissions involved some form of medical harm

External teams completed **2,374 of 2,400 planned reviews (98.9%)** and identified **429 harms** during 10,675 patient-days — a rate of **40.2 harms per 1,000 patient-days** (95% CI, 36.5 to 44.2). External reviewers consistently identified fewer harms than internal reviewers, likely reflecting their lower sensitivity in detecting harm.

### What Types of Harm Occurred?

The 588 harms identified by internal reviewers broke down as follows:

-
**Procedure-related harms: 186** (e.g., complications from surgery or invasive procedures)
-
**Medication-related harms: 162** (e.g., adverse drug reactions, dosing errors)
-
**Hospital-acquired infections: 87** (e.g., urinary tract infections, surgical-site infections, bloodstream infections)
- **Other therapies: 59**
- **Diagnostic evaluations: 7**
- **Falls: 5**
- **Other causes: 82**

### How Severe Was the Harm?

Harms were classified using the **NCC MERP Index** (National Coordinating Council for Medication Error Reporting and Prevention), which assigns categories based on severity:

-
**Category E** — Temporary harm requiring intervention: **245 harms (41.7%)**

-
**Category F** — Temporary harm requiring initial or prolonged hospitalization: **251 harms (42.7%)**

-
**Category G** — Permanent harm: **17 harms (2.9%)**

-
**Category H** — Life-threatening harm requiring intervention to sustain life: **50 harms (8.5%)**

-
**Category I** — Harm causing or contributing to death: **14 harms (2.4%)**

Of all harms, **4.4 per 100 admissions (17.9%)** were already present when the patient was admitted (meaning they resulted from prior medical care). The remaining **20.7 per 100 admissions (82.3%)** occurred during the studied hospital stay itself.

### Breakdown by Body System

The study also categorized harms by body system. The largest categories were:

-
**Hospital-acquired infections: 110 total** (82 preventable) — including 31 urinary tract infections, 17 surgical-site infections, 11 non-ventilator pneumonias, 10 cases of sepsis/bacteremia, 9 catheter-related bloodstream infections, and 8 ventilator-associated pneumonias
-
**Other types of harm: 119 total** (73 preventable) — including 35 pressure ulcers (bedsores), 9 allergic reactions, 8 falls, and 8 catheter complications
-
**Surgical or obstetrical events: 85 total** (49 preventable) — including 18 lacerations or organ injuries, 16 unplanned returns to surgery, and 6 postoperative hemorrhages
-
**Hematologic (blood) system: 53 total** (31 preventable) — including 27 hemorrhages and 5 hematomas
-
**Renal or endocrine system: 52 total** (43 preventable) — including 20 hypoglycemia (low blood sugar) events, 6 cases of fluid overload, and 6 cases of acute renal failure
-
**Cardiovascular system: 45 total** (22 preventable) — including 24 hypotension (dangerously low blood pressure) events and 9 arrhythmias
-
**Neurologic system: 44 total** (22 preventable) — including 14 oversedation events and 4 cases of delirium or encephalopathy
-
**Respiratory system: 41 total** (27 preventable) — including 10 cases of acute respiratory failure, 7 pulmonary emboli, and 6 pneumothoraces (collapsed lungs)
-
**Gastrointestinal system: 39 total** (15 preventable) — including 12 cases of nausea or vomiting, 7 cases of ileus (bowel blockage), and 5 cases of Clostridium difficile colitis

## Preventable Harms: What Could Have Been Avoided

A central question in patient-safety research is: how much of this harm could have been prevented? Physician reviewers rated preventability on a scale from 1 ("definitely not preventable") to 4 ("definitely preventable").

Of the 588 harms identified by internal reviewers, **364 (63.1%) were rated as preventable**. This is a striking finding — the majority of harms were not unavoidable complications but rather events that might have been prevented with better care.

Among the preventable harms specifically:

-
**144 were Category E** (temporary harm requiring intervention)
-
**163 were Category F** (temporary harm requiring initial or prolonged hospitalization)
-
**13 caused permanent harm** (Category G)
-
**35 were life-threatening** (Category H)
-
**9 caused or contributed to a patient's death** (Category I)

In other words, nearly two-thirds of medical harm was potentially avoidable — and that avoidable harm included permanent injuries, life-threatening events, and deaths.

## Changes Over Time: Did Patient Safety Improve?

This was the study's central question, and the answer was sobering.

### Internal Reviewer Results (Higher-Quality Data)

Using a statistical model called Poisson regression (a method for analyzing rates over time), the researchers found **no significant change** in the overall rate of harms per 1,000 patient-days (reduction factor, **0.99 per year; 95% CI, 0.95 to 1.04; P = 0.72**).

In plain language: the reduction factor of 0.99 means the harm rate decreased by only about **1% per year** — and the confidence interval includes 1.0, meaning the change was not statistically significant. The rate of harms per 100 admissions likewise showed no significant change.

Subanalyses found no significant improvement in:

-
**Preventable harms** (reduction factor, 0.99; 95% CI, 0.93 to 1.05; P = 0.77)
-
**Higher-severity harms** (NCC MERP categories F through I)

### External Reviewer Results

External reviewers also found no significant change in the overall rate of harms per 1,000 patient-days (reduction factor, **0.97; 95% CI, 0.92 to 1.03; P = 0.33**).

However, there was one encouraging signal: the unadjusted rate of *preventable* harms identified by external reviewers declined from **23.5 harms per 1,000 patient-days in 2002 to 15.0 harms per 1,000 patient-days in 2007** — a reduction factor of 0.91 (95% CI, 0.84 to 0.994; P = 0.04), which was statistically significant.

But this finding did not hold up under rigorous adjustment. When the researchers adjusted for patient demographics, hospital service, and high-risk conditions, the reduction in preventable harms identified by external reviewers **no longer reached statistical significance** (reduction factor, 0.92; 95% CI, 0.85 to 1.00; P = 0.06). The overall external harm rate similarly showed no significant change (reduction factor, 0.98; 95% CI, 0.93 to 1.04; P = 0.47).

Equally important: the apparent reduction in preventable harms from external reviews **was not supported by the higher-quality internal reviews**, which found flat rates. Since the internal review data was more reliable by all measures, the weight of evidence points to no meaningful improvement.

## Clinical Implications: What This Means for Patients

These findings validated the concerns raised by patient-safety experts in both the United States and Europe: **harm resulting from medical care remains very common**, and a decade of safety efforts had not produced measurable improvement at the hospital level.

The study's authors offered several explanations for why improvement was not seen:

- Many safety interventions, while promising, have **not been rigorously evaluated** to prove they work in real-world settings.
- Even proven interventions have **not been reliably implemented on a large scale** — a hospital may adopt a policy on paper but fail to carry it out consistently at the bedside.
- Patient-safety efforts have often focused on specific, narrow targets (e.g., reducing catheter-related infections) rather than **comprehensively addressing all types of harm**.

The study was powered to detect a substantial reduction in harm (from 40 to 30 harms per 100 admissions). It likely would have detected a large improvement if one had occurred. The fact that none was found — in a state with high engagement in safety work — is deeply concerning.

For patients, the takeaway is that hospital care carries real risks, and those risks did not measurably decrease during the study period. The study found that **approximately 1 in 5 admissions involved some form of harm**, and **nearly two-thirds of that harm was potentially preventable**.

## Study Limitations: What This Research Couldn't Prove

No study is perfect, and the authors acknowledged several important limitations:

-
**Single state:** The study was conducted only in North Carolina. While the state was chosen because it was likely to show improvement if any existed, results may not generalize to all U.S. hospitals.
-
**Retrospective design:** The study relied on reviewing medical records after the fact. Some harms may not have been documented in the charts, and the reviewers' ability to detect harm depended on the quality and completeness of those records.
-
**Possible missed harms:** The trigger tool, while validated, does not capture every possible harm. It's possible that some harms were missed, meaning the true rate of harm could be even higher than reported.
-
**Statistical limitations:** The study had 80% power to detect a reduction from 40 to 30 harms per 100 admissions over the study period. Smaller but real improvements might have gone undetected — though the near-flat trends in the internal review data suggest any such improvements were modest at best.
-
**No direct measurement of safety interventions:** The study measured harm rates but did not track which specific safety interventions each hospital implemented. It cannot link specific efforts to outcomes.

## Recommendations for Patients and Families

Given that medical harm remains common — and much of it is potentially preventable — patients and families should be proactive partners in their own safety. Based on this study and related patient-safety research, here are practical steps to consider:

1.
**Ask about your medications.** Medication-related harms were the second most common type (162 of 588 harms). Ask what each medication is for, the dose, and potential side effects. Ensure hospital staff check your allergy history and that all providers know your complete medication list.
1.
**Speak up about falls.** Falls caused harm in this study. Ask for help getting out of bed if you feel weak, keep call buttons within reach, and wear non-slip footwear.
1.
**Ask about infection prevention.** Hospital-acquired infections were the third most common harm (87 of 588, with urinary tract infections and surgical-site infections the most frequent). Ask whether catheters are necessary and when they will be removed. Remind staff and visitors to wash their hands. Ask about steps to prevent surgical-site infections if you're having surgery.
1.
**Know your risk after procedures.** Procedure-related harms were the most common type (186 of 588). Ask about the risks of any procedure, what complications might occur, and what symptoms should prompt you to call for help afterward.
1.
**Monitor for low blood sugar.** Hypoglycemia (low blood sugar) was a notable preventable harm (19 of 20 such events were rated preventable). If you have diabetes, ensure your blood sugar is monitored regularly, and know the symptoms of hypoglycemia.
1.
**Watch for pressure ulcers (bedsores).** These were among the most preventable harms (34 of 35). Ask nursing staff to help you change position regularly if you are confined to bed, and report any areas of persistent redness or skin breakdown.
1.
**Designate a family advocate.** A family member or friend who can be present during rounds, ask questions, and help track information can be a powerful safety net.

The study's authors emphasized that **further efforts are needed to translate effective safety interventions into routine practice** and to monitor health care safety over time. For now, being an informed, engaged patient remains one of the best defenses against medical harm.

## Frequently Asked Questions

### How common is medical harm during a hospital stay?

The study found 25.1 harms per 100 admissions, meaning about one in five admissions involved some form of harm. Procedure-related harms, medication-related harms, and hospital-acquired infections were the most common types. Some harms were temporary, but others caused permanent injury, life-threatening events, or contributed to death.

### Did patient safety improve over the six years of the study?

No significant improvement occurred. Internal reviewers found no significant change in overall harm rates from 2002 to 2007. External reviewers saw a possible reduction in preventable harms, but it was not statistically significant after adjustment. The study concludes that safety efforts did not measurably reduce harm during that period.

### What can patients do to reduce their risk of harm?

Patients can ask about each medication, its dose and side effects, and ensure allergy histories are checked. They can ask for help getting out of bed to avoid falls, ask whether catheters are necessary, monitor blood sugar if diabetic, and request regular position changes to prevent pressure ulcers. Having a family advocate can also help.

### What were the limitations of this study?

The study was conducted only in North Carolina, so results may not apply everywhere. It relied on medical records, which might have missed some harms. The trigger tool does not capture every possible harm. The study could only detect large reductions in harm, so smaller improvements might have gone unnoticed.

### Should I get a second opinion before hospital treatment if medical errors are common?

Yes, a second opinion can be valuable because hospital care carries real risks. In a study of over 2,300 admissions, about 1 in 5 involved harm, and nearly two-thirds of that harm was potentially preventable. Common harms included medication errors, infections, and procedure-related complications. A second opinion can help you understand your treatment options, confirm the necessity of procedures, and identify steps to reduce your risk. Diagnostic Detectives Network provides independent expert second opinions.

## Source Information

**Original article title:** Landrigan NEJM Hospital safety Study 2010

**Authors:** Christopher P. Landrigan, M.D., M.P.H., Gareth J. Parry, Ph.D., Catherine B. Bones, M.S.W., Andrew D. Hackbarth, M.Phil., Donald A. Goldmann, M.D., and Paul J. Sharek, M.D., M.P.H.

**Journal:** *New England Journal of Medicine*, 2010; Vol. 363, pages 2124–2134. Published November 25, 2010.

**Funding:** The study was funded by the Rx Foundation, which had no role in the design, data collection, analysis, or manuscript approval.

*This patient-friendly article is based on peer-reviewed research. It is intended to help patients understand the study's findings and is not a substitute for professional medical advice. If you have questions about your care, consult your healthcare provider.*


---
Publisher: Diagnostic Detectives Network (https://diagnosticdetectives.com) — independent multi-expert medical second opinions, worldwide, private-pay.
Author byline: Anton Titov, MD, PhD. Contact: https://diagnosticdetectives.com/pages/contact
Canonical page: https://diagnosticdetectives.com/products/patient-harm-in-hospitals-why-a-decade-of-safety-efforts-hasnt-reduced-medical-errors