Accuracy of COVID-19-Like Illness Diagnoses in Electronic Health Record Data: Retrospective Cohort Study

BACKGROUND

Electronic health record (EHR) data provide a unique opportunity to study the epidemiology of COVID-19, clinical outcomes of the infection, comparative effectiveness of therapies, and vaccine effectiveness but require a well-defined computable phenotype of COVID-19-like illness (CLI).

OBJECTIVE

The objective of this study was to evaluate the performance of pathogen-specific and other acute respiratory illness (ARI) International Statistical Classification of Diseases-9 and -10 codes in identifying COVID-19 cases in emergency department (ED) or urgent care (UC) and inpatient settings.

METHODS

We conducted a retrospective observational cohort study using EHR, claims, and laboratory information system data of ED or UC and inpatient encounters from 4 health systems in the United States. Patients who were aged ≥18 years, had an ED or UC or inpatient encounter for an ARI, and underwent a SARS-CoV-2 polymerase chain reaction test between March 1, 2020, and March 31, 2021, were included. We evaluated various CLI definitions using combinations of International Statistical Classification of Diseases-10 codes as follows: COVID-19-specific codes; CLI definition used in VISION network studies; ARI signs, symptoms, and diagnosis codes only; signs and symptoms of ARI only; and random forest model definitions. We evaluated the sensitivity, specificity, positive predictive value, and negative predictive value of each CLI definition using a positive SARS-CoV-2 polymerase chain reaction test as the reference standard. We evaluated the performance of each CLI definition for distinct hospitalization and ED or UC cohorts.

RESULTS

Among 90,952 hospitalizations and 137,067 ED or UC visits, 5627 (6.19%) and 9866 (7.20%) were positive for SARS-CoV-2, respectively. COVID-19-specific codes had high sensitivity (91.6%) and specificity (99.6%) in identifying patients with SARS-CoV-2 positivity among hospitalized patients. The VISION CLI definition maintained high sensitivity (95.8%) but lowered specificity (45.5%). By contrast, signs and symptoms of ARI had low sensitivity and positive predictive value (28.9% and 11.8%, respectively) but higher specificity and negative predictive value (85.3% and 94.7%, respectively). ARI diagnoses, signs, and symptoms alone had low predictive performance. All CLI definitions had lower sensitivity for ED or UC encounters. Random forest approaches identified distinct CLI definitions with high performance for hospital encounters and moderate performance for ED or UC encounters.

CONCLUSIONS

COVID-19-specific codes have high sensitivity and specificity in identifying adults with positive SARS-CoV-2 test results. Separate combinations of COVID-19-specific codes and ARI codes enhance the utility of CLI definitions in studies using EHR data in hospital and ED or UC settings.

Relative Risks of COVID-19-Associated Hospitalizations and Clinical Outcomes by Age and Race/Ethnicity-March 2020-March 2021

Background

Limited data exist on population-based risks and risk ratios (RRs) of coronavirus disease 2019 (COVID-19)–associated hospitalizations and clinical outcomes stratified by age and race/ethnicity.

Methods

Using data from electronic health records and claims from 4 US health systems for the period March 2020–March 2021, we calculated risk and RR by age and race/ethnicity for COVID-19–associated hospitalizations and clinical outcomes among adults (≥18 years). COVID-19–associated hospitalizations were defined based on COVID-19 discharge codes or a positive severe acute respiratory syndrome coronavirus 2 result. Proportions of acute exacerbations of underlying conditions were estimated among hospitalized patients with select underlying conditions, stratified by age and race/ethnicity.

Results

Among 2.6 million adults included in the patient cohort, 6879 had COVID-19–associated hospitalizations during March 2020–March 2021 (risk: 264 per 100 000 population). Compared with younger, non-Hispanic White adults, non-Hispanic Black and Hispanic adults aged ≥65 years had the highest hospitalization risk ratios (RR, 8.6; 95% CI, 7.6–9.9; and RR, 9.3; 95% CI, 8.5–10.3, respectively). Among hospitalized adults with COVID-19 and renal disease or cardiovascular disease, the highest proportion of acute renal failure (55.5%) or congestive heart failure (43.9%) occurred in older, non-Hispanic Black patients. Among hospitalized adults with chronic lung disease or asthma, the highest proportion of respiratory failure (62.9%) or asthma exacerbation (66.7%) occurred in older, Hispanic patients.

Conclusions

During the first year of the US COVID-19 pandemic in this cohort, older non-Hispanic Black and Hispanic adults had the highest relative risks of COVID-19–associated hospitalization and adverse outcomes and, among those with select underlying conditions, the highest occurrences of acute exacerbations of underlying conditions.

Serum Progesterone Levels in Pregnant Women with Obstructive Sleep Apnea: A Case Control Study

BACKGROUND

Pregnancy is a risk factor for sleep disordered breathing, including obstructive sleep apnea (OSA). Progesterone, one of the key hormones in pregnancy, a known respiratory drive stimulant, increases ventilation and may protect against OSA. We aimed to examine the relationship between circulating progesterone and OSA, after accounting for body weight and gestational age.

METHODS

A case control study was conducted of pregnant women with OSA and those at low risk for the disorder. Cases were identified by ICD-9 code and review of medical record. Controls were identified if they scored zero (never) for snoring, apnea, and gasping on the multivariable apnea prediction index questionnaire immediately following delivery. Subjects with available stored first and/or second trimester residual serum samples were then included in this study and serum analyzed for progesterone. Raw progesterone levels were adjusted for the effects of gestational age and maternal weight.

RESULTS

Twenty-seven cases and 64 controls with available serum were identified. Women with OSA had greater maternal weight and higher rates of related comorbidities, compared to controls. Progesterone levels correlated positively with gestational age and negatively with greater weight. Progesterone levels, adjusted for gestational age and maternal weight and expressed as multiples of median (MoM), were significantly lower in OSA cases compared to controls in both the first trimester (MoM = 0.71, confidence interval [95% CI] 0.60-0.83) relative to the MoM in controls of 1.00. In the second trimester levels were also lower in OSA cases (MoM = 0.84, 95% CI 0.73-0.96) compared to the MoM of 1.00 in controls.

CONCLUSIONS

Progesterone levels, after accounting for weight and gestational age, were lower in women with OSA than controls. Progesterone may play a protective role against OSA.

Placenta-secreted circulating markers in pregnant women with obstructive sleep apnea

AIMS

Obstructive sleep apnea (OSA) is associated with placenta-mediated adverse clinical outcomes. We aimed at comparing placenta-secreted proteins, such as first and second trimester Down syndrome screening markers which have been linked to preeclampsia, and markers of angiogenesis in pregnant women with OSA, and pregnant controls at low risk for OSA.

METHODS

A case-control study of pregnant women with OSA and controls at low risk for OSA was performed. Levels of first and second trimester markers were reported as multiple of median (MoM), and adjusted for body mass index (BMI). Stored samples were tested for markers of angiogenesis and adjusted for gestational age, BMI, and chronic hypertension.

RESULTS

A total of 24 women with OSA and 166 controls had screening markers. BMI was higher in cases compared to controls, P=0.01. MoM levels of placenta associated plasma protein-A (PAPP-A) were significantly lower in cases versus controls, even after adjusting for BMI (0.52 IQR 0.48 vs. 1.01 IQR 0.63, P=0.009). The ratio of soluble vascular endothelial growth factor receptor 1 to placental growth factor was significantly higher in cases than controls, even after adjusting for confounders (4.42 IQR 2.52 vs. 2.93 IQR 2.01, P=0.009).

CONCLUSION

Circulating placenta-secreted glycoproteins and markers of angiogenesis are altered in pregnant women with OSA.

Caregiver presence and patient completion of a transitional care intervention

OBJECTIVES

To evaluate the association between family caregiver presence and patient completion of the Care Transitions Intervention (CTI), a patient activation model that provides transitional care coaching for 30 days following hospital discharge.

STUDY DESIGN

A convenience sample of 2747 fee-for-service Medicare patients recruited for the CTI during inpatient medical hospitalizations at 6 hospitals in Rhode Island between January 1, 2009 and June 31, 2011.

METHODS

As part of an effectiveness trial of the CTI, Transitions Coaches recruited patients prior to hospital discharge. When a family caregiver was present during recruitment, the patient and family caregiver were coached together or the family caregiver was coached independently.

RESULTS

We hypothesized that CTI participation would be equivalent for the 2265 coached patients without a family caregiver present at recruitment, versus the 482 patients with a family caregiver. After adjusting for significant covariates, patients with family caregivers were more than 5 times as likely to complete the intervention as patients without family caregivers (AOR = 5.48; 95% CI = 4.22-7.12). Men with family caregivers were nearly 8 times as likely to complete the intervention as men without family caregivers (AOR = 7.94; 95% CI = 5.26-11.98).

CONCLUSIONS

The inclusion of a family caregiver is associated with a greater rate of completing the CTI for post discharge coaching, particularly among men; the inclusion of a family caregiver is a feasible modification to the CTI program.

Association of dementia with early rehospitalization among Medicare beneficiaries

Preventable hospital readmissions have been recognized as indicators of hospital quality, a source of increased healthcare expenditures, and a burden for patients, families, and caregivers. Despite growth of initiatives targeting risk factors associated with potentially avoidable hospital readmissions, the impact of dementia on the likelihood of rehospitalization is poorly characterized. Therefore, the primary objective of this retrospective cohort study was to investigate whether dementia was an independent predictor of 30-day readmissions. Administrative claims data for all admissions to Rhode Island hospitals in 2009 was utilized to identify hospitalizations of Medicare fee-for-service beneficiaries with a diagnosis of Alzheimer's Disease or other dementias. Demographics, measures of comorbid disease burden, and other potential confounders were extracted from the data and the odds of 30-day readmission to any United States hospital was calculated from conditional logistic regression models. From a sample of 25,839 hospitalizations, there were 3908 index admissions of Medicare beneficiaries who fulfilled the study criteria for a dementia diagnosis. Nearly 20% of admissions (n=5133) were followed by a readmission within thirty days. Hospitalizations of beneficiaries with a dementia diagnosis were more likely to be followed by a readmission within thirty days (adjusted odds ratio (AOR) 1.18; 95% CI, 1.08, 1.29), compared to hospitalizations of those of without dementia. Controlling for discharge site of care did not attenuate the association (AOR 1.21; 95% CI, 1.10, 1.33).

Is implementation of the care transitions intervention associated with cost avoidance after hospital discharge?

BACKGROUND

Poorly-executed transitions out of the hospital contribute significant costs to the healthcare system. Several evidence-based interventions can reduce post-discharge utilization.

OBJECTIVE

To evaluate the cost avoidance associated with implementation of the Care Transitions Intervention (CTI).

DESIGN

A quasi-experimental cohort study using consecutive convenience sampling.

PATIENTS

Fee-for-service Medicare beneficiaries hospitalized from 1 January 2009 to 31 May 2011 in six Rhode Island hospitals.

INTERVENTION

The CTI is a patient-centered coaching intervention to empower individuals to better manage their health. It begins in-hospital and continues for 30 days, including one home visit and one to two phone calls.

MAIN MEASURES

We examined post-discharge total utilization and costs for patients who received coaching (intervention group), who declined or were lost to follow-up (internal control group), and who were eligible, but not approached (external control group), using propensity score matching to control for baseline differences.

KEY RESULTS

Compared to matched internal controls (N = 321), the intervention group had significantly lower utilization in the 6 months after discharge and lower mean total health care costs ($14,729 vs. $18,779, P = 0.03). The cost avoided per patient receiving the intervention was $3,752, compared to internal controls. Results for the external control group were similar. Shifting of costs to other utilization types was not observed.

CONCLUSIONS

This analysis demonstrates that the CTI generates meaningful cost avoidance for at least 6 months post-hospitalization, and also provides useful metrics to evaluate the impact and cost avoidance of hospital readmission reduction programs.

Recruiting hospitalized patients for research: how do participants differ from eligible nonparticipants?

BACKGROUND

Randomized controlled trials provide strong evidence for guidelines and interventions. Yet, much of the eligible population declines to be studied.

OBJECTIVE

To identify differences between participants and eligible nonparticipants in (1) perceived stress, (2) self-efficacy, (3) recovery expectations, (4) discussing advance directives, and (5) understanding a standard prescription label (health literacy).

DESIGN

Quasi-experimental prospective cohort study in 5 acute-care hospitals.

METHODS

We approached 295 hospital inpatients as they were being recruited for a behavioral intervention and asked them to answer 5 screening questions. We matched respondents' answers to their acceptance of the behavioral intervention and to Medicare claims and enrollment data. We used multivariate logistic regression to compare consent rates based on screening-question responses.

SETTING/PATIENTS

Hospitalized fee-for-service Medicare patients.

RESULTS

Patients were less likely to consent to the behavioral intervention when they reported feeling unable to control important things in their lives (odds ratio [OR]: 0.35, 95% confidence interval [CI]: 0.14-0.92), had low recovery expectations (OR: 0.17, 95% CI: 0.06-0.45), or were confused by any question (OR: 0.11, 95% CI: 0.05-0.24). Conversely, individuals who answered the medication question incorrectly were more likely to consent to the behavioral intervention (OR: 3.82, 95% CI: 1.12-13.03). There were no significant differences in consent for patients who reported feeling overwhelmed or reported discussing advance care planning with family members or doctors.

CONCLUSIONS

Hospitalized eligible nonparticipants differ in constructs related to perceived stress, recovery expectation, and health literacy. Recognizing such characteristics may inform strategies to improve intervention recruitment in the hospital and representation in clinical trials.

Identifying star performers: the relationship between ambitious targets and nursing home quality improvement

Setting Targets--Achieving Results (STAR) is a Web-based tool that helps nursing home leadership select annual performance goals, or targets, for a subset of publicly reported quality measures. Previous results demonstrate that nursing homes whose staff implement STAR targets demonstrate greater improvement on the related outcomes. In this analysis, the authors hypothesized that nursing homes whose staff select the most ambitious targets (reflecting large improvement over their current performance) may be more successful in their related quality improvement efforts than homes with less-ambitious targets (reflecting lesser improvement). The authors analyzed data from 7,091 Medicare- or Medicaid-certified nursing homes that set STAR targets in 2005 or 2006 for two quality measures: the proportion of residents who were physically restrained daily and the proportion of high-risk residents with pressure ulcers. Targets were classified as ambitious or less ambitious based on the 75th and 50th rank-ordered percentiles, respectively. Improvement was calculated using four-quarter averages for baseline (the year ending when the target was set) and remeasurement (the subsequent year). The results indicate that nursing homes with ambitious targets demonstrate greater improvement than their peers selecting less-ambitious targets. With limited federal and local resources to assist providers with quality improvement, target values may be a used as a "flag" to help agencies allocate scarce resources to nursing homes committed to quality improvement efforts and with the organizational capacity to improve.

Inheritance and segregation of virus and herbicide resistance transgenes in sugarcane

Transgenic sugarcane parents containing multiple copies of herbicide resistance ( bar) and Sorghum mosaic virus (SrMV) resistance ( hut) genes were crossed with non-transgenic sugarcane varieties. Segregation of the transgenes in the progeny was determined using Southern blot analysis; herbicide resistance and SrMV resistance were assessed using bioassays. The segregation data were used to infer linkage relationships between transgenes in the parent plants. In two of the parents, all transgene insertions were linked in one position in the genome, although some recombination between insertion events did occur. In the third parent, insertion had occurred in two independent, unlinked loci. Analysis of progeny of this parent indicated that rearrangement or mutation occurred in both loci, resulting in non-parental transgene DNA fragments in some progeny. Most transgenic progeny containing the bar gene showed resistance to herbicide. SrMV inoculation indicated that a fairly high proportion of the transgenic progeny showed susceptibility. As the post-transcriptional gene silencing mechanism responsible for the virus resistance phenotype may be reset during meiosis, phenotypic screening of older plants may be a more reliable indication of virus resistance than screening young seedlings. This is the first report of transgene segregation in sugarcane, and we have demonstrated that transgenic sugarcane parents showing stable inheritance of transgenes can be effectively used in breeding programs.

A study to investigate the ability of subjects with chronic lung diseases to activate the roadside Lion Alcolmeter SL-400

The Lion Alcolmeter SL-2 is widely used for road-side breath-testing by police in the UK. However, some individuals with lung diseases have difficulty in activating the device. This study describes an investigation that we have carried out on a new device called the Lion Alcolmeter SL-400 which has recently been introduced into use by police forces in the UK. The manufacturers state that the machine requires a minimum continuous expiratory flow rate of 25 litres per minute and a minimum expired volume of 1.5 litres, after which a breath sample is automatically taken and analysed. Our study was designed to investigate the ability of subjects with a variety of lung diseases to activate this device. The 40 adult subjects investigated consisted of 10 normal controls, 10 with asthma, 10 with chronic obstructive pulmonary disease and 10 with restrictive lung disease. After baseline lung function tests were performed, the subjects were then given alcohol to drink, the amount of which was based upon their body weight. After a gap of at least 20 minutes, the subjects were then asked to attempt up to three blows into the Alcolmeter. Our results showed that three asthmatic subjects, four with chronic obstructive pulmonary disease and two with restrictive lung disease, failed to successfully activate the device even after three attempts. All of the subjects who failed to activate the device had an expired breath volume of more than 1.5 litres, but seven out of these nine subjects had a Forced Expiratory Volume (FEV1) of less than 1.1 litres. In conclusion, this study has shown that some subjects with lung diseases may have difficulty in activating the SL-400 roadside alcolmeter device.

The effects of chronic obstructive airways disease on the ability to drive and to use a roadside alcolmeter

Subjects with chronic obstructive airways disease may have difficulty with the roadside alcolmeter. Twenty-six subjects with a FEV1/FVC less than 60% were asked to use an alcolmeter simulator. Only ten were able to produce the necessary flow rate of 28 l min-1 for a minimum of 2.7 s, two could produce the same total volume (1.25 l) at 10 l min-1 for 7.5 s, five could only expel 10 l min-1 for 4.5 s, and nine were unable to trigger the alcolmeter at even these very low flow rates. Subjects with an FEV1 of less than 1.51 or FEV1% predicted less than 50% were very unlikely to be able to activate the alcolmeter. Ten healthy subjects were investigated to assess the accuracy of the roadside alcolmeter at a flow rate of 10 l min-1 compared to 40 l min-1. No significant difference was found in breath alcohol levels between the two flow rates. It is proposed that some modification could be made to the roadside alcolmeter, without affecting its accuracy, to allow some subjects with chronic obstructive lung disease to activate the device. A postal survey of 284 subjects with a FEV1/FVC less than 60% was carried out. Of those who were drivers or exdrivers, 24.7% had had to stop or reduce their driving because of their respiratory disease. This group had a significantly lower FEV1% predicted (P = 0.035) than those whose driving was unaffected.

A subset of memory CD4+ helper T lymphocytes identified by expression of Pgp-1

The Pgp-1 glycoprotein (Ly-24 antigen) is acquired by mature murine T lymphocytes at the time of primary antigen stimulation Pgp-1 was previously shown to be a useful cell surface marker for distinguishing antigen-specific memory CD8+ T lymphocytes after immunization. Here we demonstrate that this observation extends to CD4+ T lymphocytes. Antigen-specific CD4+ T cells in mice immunized with sperm whale myoglobin or keyhole limpet hemocyanin were contained nearly exclusively in the minor Pgp-1+ subset.