Using natural language processing to identify the status of homelessness and housing instability among serious illness patients from clinical notes in an integrated healthcare system.

Background: Efficiently identifying the social risks of patients with serious illnesses (SIs) is the critical first step in providing patient-centered and value-driven care for this medically vulnerable population. Objective: To apply and further hone an existing natural language process (NLP) algorithm that identifies patients who are homeless/at risk of homeless to a SI population. Methods: Patients diagnosed with SI between 2019 and 2020 were identified using an adapted list of diagnosis codes from the Center for Advance Palliative Care from the Kaiser Permanente Southern California electronic health record. Clinical notes associated with medical encounters within 6 months before and after the diagnosis date were processed by a previously developed NLP algorithm to identify patients who were homeless/at risk of homelessness. To improve the generalizability to the SI population, the algorithm was refined by multiple iterations of chart review and adjudication. The updated algorithm was then applied to the SI population. Results: Among 206 993 patients with a SI diagnosis, 1737 (0.84%) were identified as homeless/at risk of homelessness. These patients were more likely to be male (51.1%), age among 45-64 years (44.7%), and have one or more emergency visit (65.8%) within a year of their diagnosis date. Validation of the updated algorithm yielded a sensitivity of 100.0% and a positive predictive value of 93.8%. Conclusions: The improved NLP algorithm effectively identified patients with SI who were homeless/at risk of homelessness and can be used to target interventions for this vulnerable group.

Chronic and Sustained High-Dose Opioid Use in an Integrated Health System.

INTRODUCTION: The Centers for Disease Control and Prevention Guideline for Prescribing Opioids for Chronic Pain released in 2016 had led to decreases in opioid prescribing. This study sought to examine chronic and sustained high-dose prescription opioid use in an integrated health system. METHODS: A serial cross-sectional study was conducted in 2021 to estimate the annual age-adjusted prevalence and incidence of chronic and high-dose opioid use among demographically diverse noncancer adults in an integrated health system in Southern California during 2013-2020. Interrupted time-series analysis with segmented regression was conducted to estimate changes in the trends in annual rates before (2013-2015) and after (2017-2020) the 2016 guideline, treating 2016 as a wash-out period. RESULTS: Prevalence and incidence of chronic use and sustained high-dose use had started to decrease after a health system intervention program before the 2016 Centers for Disease Control and Prevention guideline release and continued to decline after the guideline. Among those with sustained high-dose use, there was a substantial decrease in persons with an average daily dosage ≥90 morphine milligram equivalent and concurrent benzodiazepine use. An accelerated decrease in prevalent chronic use after the guideline was observed (slope change: -11.1 [95% CI= -20.3, -1.9] users/10,000 person-years, p=0.03). The incidence of chronic use and sustained high-dose use continued to decrease after the guideline release but at a slower pace. CONCLUSIONS: Implementing evidence-based prescribing guidelines was associated with a decrease in chronic and sustained high-dose prescription opioid use.

Analysis of mRNA COVID-19 Vaccine Uptake Among Immunocompromised Individuals in a Large US Health System.

Importance: Immunocompromised individuals are at increased risk for severe outcomes due to SARS-CoV-2 infection. Given the varying and complex nature of COVID-19 vaccination recommendations, it is important to understand COVID-19 vaccine uptake in this vulnerable population. Objective: To assess mRNA COVID-19 vaccine uptake and factors associated with uptake among immunocompromised individuals from December 14, 2020, through August 6, 2022. Design, Setting, and Participants: This cohort study was conducted with patients of Kaiser Permanente Southern California (KPSC), an integrated health care system in the US. The study included patients aged 18 years or older who were immunocompromised (individuals with an immunocompromising condition or patients who received immunosuppressive medications in the year prior to December 14, 2020) and still met criteria for being immunocompromised 1 year later. Exposures: Age, sex, self-identified race and ethnicity, prior positive COVID-19 test result, immunocompromising condition, immunomodulating medication, comorbidities, health care utilization, and neighborhood median income. Main Outcomes and Measures: Outcomes were the number of doses of mRNA COVID-19 vaccine received and the factors associated with receipt of at least 4 doses, estimated by hazard ratios (HRs) and 95% Wald CIs via Cox proportional hazards regression. Statistical analyses were conducted between August 9 and 23, 2022. Results: Overall, 42 697 immunocompromised individuals met the study eligibility criteria. Among these, 18 789 (44.0%) were aged 65 years or older; 20 061 (47.0%) were women and 22 635 (53.0%) were men. With regard to race and ethnicity, 4295 participants (10.1%) identified as Asian or Pacific Islander, 5174 (12.1%) as Black, 14 289 (33.5%) as Hispanic, and 17 902 (41.9%) as White. As of the end of the study period and after accounting for participant censoring due to death or disenrollment from the KPSC health plan, 78.0% of immunocompromised individuals had received a third dose of mRNA COVID-19 vaccine. Only 41.0% had received a fourth dose, which corresponds to a primary series and a monovalent booster dose for immunocompromised individuals. Uptake of a fifth dose was only 0.9% following the US Centers for Disease Control and Prevention (CDC) recommendation to receive a second monovalent booster (ie, fifth dose). Adults aged 65 years or older (HR, 3.95 [95% CI, 3.70-4.22]) were more likely to receive at least 4 doses compared with those aged 18 to 44 years or 45 to 64 years (2.52 [2.36-2.69]). Hispanic and non-Hispanic Black adults (HR, 0.77 [95% CI, 0.74-0.80] and 0.82 [0.78-0.87], respectively, compared with non-Hispanic White adults), individuals with prior documented SARS-CoV-2 infection (0.71 [0.62-0.81] compared with those without), and individuals receiving high-dose corticosteroids (0.88 [0.81-0.95] compared with those who were not) were less likely to receive at least 4 doses. Conclusions and Relevance: These findings suggest that adherence to CDC mRNA monovalent COVID-19 booster dose recommendations among immunocompromised individuals was low. Given the increased risk for severe COVID-19 in this vulnerable population and the well-established additional protection afforded by booster doses, targeted and tailored efforts to ensure that immunocompromised individuals remain up to date with COVID-19 booster dose recommendations are warranted.

Demographic, Reproductive, and Medical Risk Factors for Intrauterine Device Expulsion.

OBJECTIVE: To explore to what extent intrauterine device (IUD) expulsion is associated with demographic and clinical risk factors. METHODS: The APEX-IUD (Association of Perforation and Expulsion of IntraUterine Devices) study was a U.S. cohort study using electronic health records from three integrated health care systems (Kaiser Permanente Northern California, Southern California, and Washington) and a health care information exchange (Regenstrief Institute). These analyses included individuals aged 50 years or younger with IUD insertions from 2001 to 2018. Intrauterine device expulsion cumulative incidence and incidence rates were estimated. Using Cox regression models, hazard ratios with 95% CIs were estimated before and after adjustment for risk factors of interest (age, race and ethnicity, parity, body mass index [BMI], heavy menstrual bleeding, and dysmenorrhea) and potential confounders. RESULTS: In total, 228,834 individuals with IUD insertion and no delivery in the previous 52 weeks were identified (184,733 [80.7%] with levonorgestrel-releasing intrauterine system). Diagnosis of heavy menstrual bleeding-particularly a diagnosis in both recent and past periods-was the strongest risk factor for IUD expulsion. Categories with the highest risk of IUD expulsion within each risk factor included individuals diagnosed with overweight, obesity, and morbid obesity; those in younger age groups, especially among those aged 24 years or younger; and in those with parity of four or more. Non-Hispanic White individuals had the lowest incidence and risk, and after adjustment, Asian or Pacific Islander individuals had the highest risk. Dysmenorrhea was not independently associated with expulsion risk when adjusting for heavy menstrual bleeding. CONCLUSION: Most risk factors for expulsion identified in this study appear consistent with known physiologic factors that affect uterine anatomy and physiology (age, BMI, heavy menstrual bleeding, parity). The increased risk of IUD expulsion among individuals of color warrants further investigation. Intrauterine devices are an effective long-term contraceptive; expulsion is uncommon, but patients should be counseled accordingly. FUNDING SOURCE: Bayer AG. CLINICAL TRIAL REGISTRATION: EU PAS register, EUPAS33461.

Postpartum Migraine Headache Coding in Electronic Health Records of a Large Integrated Health Care System: Validation Study.

BACKGROUND: Migraine is a common neurological disorder characterized by repeated headaches of varying intensity. The prevalence and severity of migraine headaches disproportionally affects women, particularly during the postpartum period. Moreover, migraines during pregnancy have been associated with adverse maternal outcomes, including preeclampsia and postpartum stroke. However, due to the lack of a validated instrument for uniform case ascertainment on postpartum migraine headache, there is uncertainty in the reported prevalence in the literature. OBJECTIVE: The aim of this study was to evaluate the completeness and accuracy of reporting postpartum migraine headache coding in a large integrated health care system's electronic health records (EHRs) and to compare the coding quality before and after the implementation of the International Classification of Diseases, 10th revision, Clinical Modification (ICD-10-CM) codes and pharmacy records in EHRs. METHODS: Medical records of 200 deliveries in all 15 Kaiser Permanente Southern California hospitals during 2 time periods, that is, January 1, 2012 through December 31, 2014 (International Classification of Diseases, 9th revision, Clinical Modification [ICD-9-CM] coding period) and January 1, 2017 through December 31, 2019 (ICD-10-CM coding period), were randomly selected from EHRs for chart review. Two trained research associates reviewed the EHRs for all 200 women for postpartum migraine headache cases documented within 1 year after delivery. Women were considered to have postpartum migraine headache if either a mention of migraine headache (yes for diagnosis) or a prescription for treatment of migraine headache (yes for pharmacy records) was noted in the electronic chart. Results from the chart abstraction served as the gold standard and were compared with corresponding diagnosis and pharmacy prescription utilization records for both ICD-9-CM and ICD-10-CM coding periods through comparisons of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), as well as the summary statistics of F-score and Youden J statistic (J). The kappa statistic (κ) for interrater reliability was calculated. RESULTS: The overall agreement between the identification of migraine headache using diagnosis codes and pharmacy records compared to the medical record review was strong. Diagnosis coding (F-score=87.8%; J=82.5%) did better than pharmacy records (F-score=72.7%; J=57.5%) when identifying cases, but combining both of these sources of data produced much greater accuracy in the identification of postpartum migraine cases (F-score=96.9%; J=99.7%) with sensitivity, specificity, PPV, and NPV of 100%, 99.7%, 93.9%, and 100%, respectively. Results were similar across the ICD-9-CM (F-score=98.7%, J=99.9%) and ICD-10-CM coding periods (F-score=94.9%; J=99.6%). The interrater reliability between the 2 research associates for postpartum migraine headache was 100%. CONCLUSIONS: Neither diagnostic codes nor pharmacy records alone are sufficient for identifying postpartum migraine cases reliably, but when used together, they are quite reliable. The completeness of the data remained similar after the implementation of the ICD-10-CM coding in the EHR system.

Natural Language Processing for Improved Characterization of COVID-19 Symptoms: Observational Study of 350,000 Patients in a Large Integrated Health Care System.

BACKGROUND: Natural language processing (NLP) of unstructured text from electronic medical records (EMR) can improve the characterization of COVID-19 signs and symptoms, but large-scale studies demonstrating the real-world application and validation of NLP for this purpose are limited. OBJECTIVE: The aim of this paper is to assess the contribution of NLP when identifying COVID-19 signs and symptoms from EMR. METHODS: This study was conducted in Kaiser Permanente Southern California, a large integrated health care system using data from all patients with positive SARS-CoV-2 laboratory tests from March 2020 to May 2021. An NLP algorithm was developed to extract free text from EMR on 12 established signs and symptoms of COVID-19, including fever, cough, headache, fatigue, dyspnea, chills, sore throat, myalgia, anosmia, diarrhea, vomiting or nausea, and abdominal pain. The proportion of patients reporting each symptom and the corresponding onset dates were described before and after supplementing structured EMR data with NLP-extracted signs and symptoms. A random sample of 100 chart-reviewed and adjudicated SARS-CoV-2-positive cases were used to validate the algorithm performance. RESULTS: A total of 359,938 patients (mean age 40.4 [SD 19.2] years; 191,630/359,938, 53% female) with confirmed SARS-CoV-2 infection were identified over the study period. The most common signs and symptoms identified through NLP-supplemented analyses were cough (220,631/359,938, 61%), fever (185,618/359,938, 52%), myalgia (153,042/359,938, 43%), and headache (144,705/359,938, 40%). The NLP algorithm identified an additional 55,568 (15%) symptomatic cases that were previously defined as asymptomatic using structured data alone. The proportion of additional cases with each selected symptom identified in NLP-supplemented analysis varied across the selected symptoms, from 29% (63,742/220,631) of all records for cough to 64% (38,884/60,865) of all records with nausea or vomiting. Of the 295,305 symptomatic patients, the median time from symptom onset to testing was 3 days using structured data alone, whereas the NLP algorithm identified signs or symptoms approximately 1 day earlier. When validated against chart-reviewed cases, the NLP algorithm successfully identified signs and symptoms with consistently high sensitivity (ranging from 87% to 100%) and specificity (94% to 100%). CONCLUSIONS: These findings demonstrate that NLP can identify and characterize a broad set of COVID-19 signs and symptoms from unstructured EMR data with enhanced detail and timeliness compared with structured data alone.

Cohort profile: pathways to care among people with disorders of sex development (DSD).

PURPOSE: The 'DSD Pathways' study was initiated to assess health status and patterns of care among people enrolled in large integrated healthcare systems and diagnosed with conditions comprising the broad category of disorders (differences) of sex development (DSD). The objectives of this communication are to describe methods of cohort ascertainment for two specific DSD conditions-classic congenital adrenal hyperplasia with 46,XX karyotype (46,XX CAH) and complete androgen insensitivity syndrome (CAIS). PARTICIPANTS: Using electronic health records we developed an algorithm that combined diagnostic codes, clinical notes, laboratory data and pharmacy records to assign each cohort candidate a 'strength-of-evidence' score supporting the diagnosis of interest. A sample of cohort candidates underwent a review of the full medical record to determine the score cutoffs for final cohort validation. FINDINGS TO DATE: Among 5404 classic 46,XX CAH cohort candidates the strength-of-evidence scores ranged between 0 and 10. Based on sample validation, the eligibility cut-off for full review was set at the strength-of-evidence score of ≥7 among children under the age of 8 years and ≥8 among older cohort candidates. The final validation of all cohort candidates who met the cut-off criteria identified 115 persons with classic 46,XX CAH. The strength-of-evidence scores among 648 CAIS cohort candidates ranged from 2 to 10. There were no confirmed CAIS cases among cohort candidates with scores <6. The in-depth medical record review for candidates with scores ≥6 identified 61 confirmed cases of CAIS. FUTURE PLANS: As the first cohort of this type, the DSD Pathways study is well-positioned to fill existing knowledge gaps related to management and outcomes in this heterogeneous population. Analyses will examine diagnostic and referral patterns, adherence to care recommendations and physical and mental health morbidities examined through comparisons of DSD and reference populations and analyses of health status across DSD categories.

Identification of Preterm Labor Evaluation Visits and Extraction of Cervical Length Measures from Electronic Health Records Within a Large Integrated Health Care System: Algorithm Development and Validation.

BACKGROUND: Preterm birth (PTB) represents a significant public health problem in the United States and throughout the world. Accurate identification of preterm labor (PTL) evaluation visits is the first step in conducting PTB-related research. OBJECTIVE: We aimed to develop a validated computerized algorithm to identify PTL evaluation visits and extract cervical length (CL) measures from electronic health records (EHRs) within a large integrated health care system. METHODS: We used data extracted from the EHRs at Kaiser Permanente Southern California between 2009 and 2020. First, we identified triage and hospital encounters with fetal fibronectin (fFN) tests, transvaginal ultrasound (TVUS) procedures, PTL medications, or PTL diagnosis codes within 240/7-346/7 gestational weeks. Second, clinical notes associated with triage and hospital encounters within 240/7-346/7 gestational weeks were extracted from EHRs. A computerized algorithm and an automated process were developed and refined by multiple iterations of chart review and adjudication to search the following PTL indicators: fFN tests, TVUS procedures, abdominal pain, uterine contractions, PTL medications, and descriptions of PTL evaluations. An additional process was constructed to extract the CLs from the corresponding clinical notes of these identified PTL evaluation visits. RESULTS: A total of 441,673 live birth pregnancies were identified between 2009 and 2020. Of these, 103,139 pregnancies (23.35%) had documented PTL evaluation visits identified by the computerized algorithm. The trend of pregnancies with PTL evaluation visits slightly decreased from 24.41% (2009) to 17.42% (2020). Of the first 103,139 PTL visits, 19,439 (18.85%) and 44,423 (43.97%) had an fFN test and a TVUS, respectively. The percentage of first PTL visits with an fFN test decreased from 18.06% at 240/7 gestational weeks to 2.32% at 346/7 gestational weeks, and TVUS from 54.67% at 240/7 gestational weeks to 12.05% in 346/7 gestational weeks. The mean (SD) of the CL was 3.66 (0.99) cm with a mean range of 3.61-3.69 cm that remained stable across the study period. Of the pregnancies with PTL evaluation visits, the rate of PTB remained stable over time (20,399, 19.78%). Validation of the computerized algorithms against 100 randomly selected records from these potential PTL visits showed positive predictive values of 97%, 94.44%, 100%, and 96.43% for the PTL evaluation visits, fFN tests, TVUS, and CL, respectively, along with sensitivity values of 100%, 90%, and 90%, and specificity values of 98.8%, 100%, and 98.6% for the fFN test, TVUS, and CL, respectively. CONCLUSIONS: The developed computerized algorithm effectively identified PTL evaluation visits and extracted the corresponding CL measures from the EHRs. Validation against this algorithm achieved a high level of accuracy. This computerized algorithm can be used for conducting PTL- or PTB-related pharmacoepidemiologic studies and patient care reviews.

Effectiveness Associated With BNT162b2 Vaccine Against Emergency Department and Urgent Care Encounters for Delta and Omicron SARS-CoV-2 Infection Among Adolescents Aged 12 to 17 Years.

Importance: Data about the duration of protection of 2 and 3 doses of BNT162b2 in children and adolescents are needed to help inform recommendations for boosters in this age group. Objective: To evaluate vaccine effectiveness (VE) and durability associated with 2 doses of BNT162b2 against Delta- and Omicron-related emergency department (ED) and urgent care (UC) encounters among adolescents aged 12 to 17 years and to estimate VE associated with 3 doses against these same outcomes. Design, Setting, and Participants: This test-negative case-control study was conducted at Kaiser Permanente Southern California, an integrated health care system using electronic health records in the US. Participants included Kaiser Permanente Southern California members ages 12 to 17 years with an ED or UC encounter from November 1, 2021, through March 18, 2022, for acute respiratory infection who were tested for SARS-CoV-2 via a reverse transction-polymerase chain reaction test. Analyses were conducted from March 21 to June 22, 2022. Exposures: BNT162b2 vaccination status ascertained from electronic health records and state registry data. Main Outcomes and Measures: The main outcome was VE associated with BNT162b2 against ED and UC encounters related to Delta or Omicron variant SARS-CoV-2 infection. Results: Analyses were conducted among 3168 adolescents, including 1004 with ED visits and 2164 with UC visits. Median (IQR) age was 15 (13-16) years, and 1461 (46.1%) were boys. In adjusted analyses, VE associated with 2 doses of BNT162b2 against ED or UC encounters was highest within the first 2 months for both Delta (89% [95% CI, 69% to 96%]) and Omicron (73% [95% CI, 54% to 84%]) variants but waned to 49% (95% CI, 27% to 65%) for the Delta variant and 16% (95% CI, -7% to 34%) for the Omicron variant at 6 months and beyond. A third dose of BNT162b2 was associated with improved protection against the Omicron variant (87% [95% CI, 72% to 94%]) after a median (IQR) of 19 (9-32) days after dose 3. Conclusions and Relevance: These findings suggest that 2 doses of the BNT162b2 COVID-19 vaccine were associated with high levels of protection against ED and UC encounters related to the Delta and Omicron variants of SARS-CoV-2 in the first few months after vaccination. However, effectiveness waned over time, especially against Omicron. A third dose of BNT162b2 was associated with improved protection against Omicron beyond that seen initially after 2 doses, underscoring the importance of boosters for adolescents aged 12 to 17 years.

Identifying Suicidal Ideation and Attempt From Clinical Notes Within a Large Integrated Health Care System.

PURPOSE: The purpose of this study was to develop a natural language processing algorithm to identify suicidal ideation/attempt from free-text clinical notes. METHODS: Clinical notes containing prespecified keywords related to suicidal ideation/attempts from 2010 to 2018 were extracted from our organization's electronic health record system. A random sample of 864 clinical notes was selected and equally divided into 4 subsets. These subsets were reviewed and classified as 1 of the following 3 suicidal ideation/attempt categories (current, historical, and no) by experienced research chart abstractors. The first 3 data sets were used to develop the rule-based computerized algorithm sequentially and the fourth data set was used to evaluate the algorithm's performance. The validated algorithm was then applied to the entire study sample of clinical notes. RESULTS: The computerized algorithm correctly identified 23 of the 26 confirmed current suicidal ideation/attempts and all 10 confirmed historical suicidal ideation/attempts in the validation data set. It produced an 88.5% sensitivity and a 100.0% positive predictive value for current suicidal ideation/attempts, and a 100.0% sensitivity and positive predictive value for historical suicidal ideation/attempts. After applying the computerized algorithm to the entire set of study notes, we identified a total of 1,050,287 current ideation/attempt events and 293,037 historical ideation/attempt events documented in clinical notes. Those for which current ideation/attempt events were documented were more likely to be female (59.5%), 25-44 years old (28.3%), and White (43.4%). CONCLUSION: Our study demonstrated that a computerized algorithm can effectively identify suicidal ideation/attempts from clinical notes. This algorithm can be utilized in support of suicide prevention research programs and patient care quality improvement initiatives.

Association between intrauterine device type and risk of perforation and device expulsion: results from the Association of Perforation and Expulsion of Intrauterine Device study.

BACKGROUND: Intrauterine devices, including levonorgestrel-releasing and copper devices, are highly effective long-acting reversible contraceptives. The potential risks associated with intrauterine devices are low and include uterine perforation and device expulsion. OBJECTIVE: This study aimed to evaluate the risk of perforation and expulsion associated with levonorgestrel-releasing devices vs copper devices in clinical practice in the United States. STUDY DESIGN: The Association of Perforation and Expulsion of Intrauterine Device study was a retrospective cohort study of women aged ≤50 years with an intrauterine device insertion during 2001 to 2018 and information on intrauterine device type and patient and medical characteristics. Of note, 4 research sites with access to electronic health records contributed data for the study: 3 Kaiser Permanente-integrated healthcare systems (Northern California, Southern California, and Washington) and 1 healthcare system using data from a healthcare information exchange in Indiana (Regenstrief Institute). Perforation was classified as any extension of the device into or through the myometrium. Expulsion was classified as complete (not visible in the uterus or abdomen or patient reported) or partial (any portion in the cervix or malpositioned). We estimated the crude incidence rates and crude cumulative incidence by intrauterine device type. The risks of perforation and expulsion associated with levonorgestrel-releasing intrauterine devices vs copper intrauterine devices were estimated using Cox proportional-hazards regression with propensity score overlap weighting to adjust for confounders. RESULTS: Among 322,898 women included in this analysis, the incidence rates of perforation per 1000 person-years were 1.64 (95% confidence interval, 1.53-1.76) for levonorgestrel-releasing intrauterine devices and 1.27 (95% confidence interval, 1.08-1.48) for copper intrauterine devices; 1-year and 5-year crude cumulative incidence was 0.22% (95% confidence interval, 0.20-0.24) and 0.63% (95% confidence interval, 0.57-0.68) for levonorgestrel-releasing intrauterine devices and 0.16% (95% confidence interval, 0.13-0.20) and 0.55% (95% confidence interval, 0.44-0.68) for copper intrauterine devices, respectively. The incidence rates of expulsion per 1000 person-years were 13.95 (95% confidence interval, 13.63-14.28) for levonorgestrel-releasing intrauterine devices and 14.08 (95% confidence interval, 13.44-14.75) for copper intrauterine devices; 1-year and 5-year crude cumulative incidence was 2.30% (95% confidence interval, 2.24-2.36) and 4.52% (95% confidence interval, 4.40-4.65) for levonorgestrel-releasing intrauterine devices and 2.30% (95% confidence interval, 2.18-2.44) and 4.82 (95% confidence interval, 4.56-5.10) for copper intrauterine devices, respectively. Comparing levonorgestrel-releasing intrauterine devices with copper intrauterine devices, the adjusted hazard ratios were 1.49 (95% confidence intervals, 1.25-1.78) for perforation and 0.69 (95% confidence intervals, 0.65-0.73) for expulsion. CONCLUSION: After adjusting for potential confounders, levonorgestrel-releasing intrauterine devices were associated with an increased risk of uterine perforation and a decreased risk of expulsion relative to copper intrauterine devices. Given that the absolute numbers of these events are low in both groups, these differences may not be clinically meaningful.

Durability of BNT162b2 vaccine against hospital and emergency department admissions due to the omicron and delta variants in a large health system in the USA: a test-negative case-control study.

BACKGROUND: The duration of protection against the omicron (B.1.1.529) variant for current COVID-19 vaccines is not well characterised. Vaccine-specific estimates are especially needed. We aimed to evaluate the effectiveness and durability of two and three doses of the BNT162b2 (Pfizer-BioNTech) mRNA vaccine against hospital and emergency department admissions due to the delta (B.1.617.2) and omicron variants. METHODS: In this case-control study with a test-negative design, we analysed electronic health records of members of Kaiser Permanente Southern California (KPSC), a large integrated health system in California, USA, from Dec 1, 2021, to Feb 6, 2022. Vaccine effectiveness was calculated in KPSC patients aged 18 years and older admitted to hospital or an emergency department (without a subsequent hospital admission) with a diagnosis of acute respiratory infection and tested for SARS-CoV-2 via PCR. Adjusted vaccine effectiveness was estimated with odds ratios from adjusted logistic regression models. This study is registered with ClinicalTrials.gov (NCT04848584). FINDINGS: Analyses were done for 11 123 hospital or emergency department admissions. In adjusted analyses, effectiveness of two doses of the BNT162b2 vaccine against the omicron variant was 41% (95% CI 21-55) against hospital admission and 31% (16-43) against emergency department admission at 9 months or longer after the second dose. After three doses, effectiveness of BNT162b2 against hospital admission due to the omicron variant was 85% (95% CI 80-89) at less than 3 months but fell to 55% (28-71) at 3 months or longer, although confidence intervals were wide for the latter estimate. Against emergency department admission, the effectiveness of three doses of BNT162b2 against the omicron variant was 77% (72-81) at less than 3 months but fell to 53% (36-66) at 3 months or longer. Trends in waning against SARS-CoV-2 outcomes due to the delta variant were generally similar, but with higher effectiveness estimates at each timepoint than those seen for the omicron variant. INTERPRETATION: Three doses of BNT162b2 conferred high protection against hospital and emergency department admission due to both the delta and omicron variants in the first 3 months after vaccination. However, 3 months after receipt of a third dose, waning was apparent against SARS-CoV-2 outcomes due to the omicron variant, including hospital admission. Additional doses of current, adapted, or novel COVD-19 vaccines might be needed to maintain high levels of protection against subsequent waves of SARS-CoV-2 caused by the omicron variant or future variants with similar escape potential. FUNDING: Pfizer.

Association between menorrhagia and risk of intrauterine device-related uterine perforation and device expulsion: results from the Association of Uterine Perforation and Expulsion of Intrauterine Device study.

BACKGROUND: Intrauterine devices are effective instruments for contraception, and 1 levonorgestrel-releasing device is also indicated for the treatment of heavy menstrual bleeding (menorrhagia). OBJECTIVE: To compare the incidence of intrauterine device expulsion and uterine perforation in women with and without a diagnosis of menorrhagia within the first 12 months before device insertion STUDY DESIGN: This was a retrospective cohort study conducted in 3 integrated healthcare systems (Kaiser Permanente Northern California, Southern California, and Washington) and a healthcare information exchange (Regenstrief Institute) in the United States using electronic health records. Nonpostpartum women aged ≤50 years with intrauterine device (eg, levonorgestrel or copper) insertions from 2001 to 2018 and without a delivery in the previous 12 months were studied in this analysis. Recent menorrhagia diagnosis (ie, recorded ≤12 months before insertion) was ascertained from the International Classification of Diseases, Ninth and Tenth Revision, Clinical Modification codes. The study outcomes, viz, device expulsion and device-related uterine perforation (complete or partial), were ascertained from electronic medical records and validated in the data sources. The cumulative incidence and crude incidence rates with 95% confidence intervals were estimated. Cox proportional hazards models estimated the crude and adjusted hazard ratios using propensity score overlap weighting (13-16 variables) and 95% confidence intervals. RESULTS: Among 228,834 nonpostpartum women, the mean age was 33.1 years, 44.4% of them were White, and 31,600 (13.8%) had a recent menorrhagia diagnosis. Most women had a levonorgestrel-releasing device (96.4% of those with and 78.2% of those without a menorrhagia diagnosis). Women with a menorrhagia diagnosis were likely to be older, obese, and have dysmenorrhea or fibroids. Women with a menorrhagia diagnosis had a higher intrauterine device-expulsion rate (40.01 vs 10.92 per 1000 person-years) than those without, especially evident in the first few months after insertion. Women with a menorrhagia diagnosis had a higher cumulative incidence (95% confidence interval) of expulsion (7.00% [6.70-7.32] at 1 year and 12.03% [11.52-12.55] at 5 years) vs those without (1.77% [1.70-1.84] at 1 year and 3.69% [3.56-3.83] at 5 years). The risk of expulsion was increased for women with a menorrhagia diagnosis vs for those without (adjusted hazard ratio, 2.84 [95% confidence interval, 2.66-3.03]). The perforation rate was low overall (<1/1000 person-years) but higher in women with a diagnosis of menorrhagia vs in those without (0.98 vs 0.63 per 1000 person-years). The cumulative incidence (95% confidence interval) of uterine perforation was slightly higher for women with a menorrhagia diagnosis (0.09% [0.06-0.14] at 1 year and 0.39% [0.29-0.53] at 5 years) than those without it (0.07% [0.06-0.08] at 1 year and 0.28% [0.24-0.33] at 5 years). The risk of perforation was slightly increased in women with a menorrhagia diagnosis vs in those without (adjusted hazard ratio, 1.53; 95% confidence interval, 1.10-2.13). CONCLUSION: The risk of expulsion is significantly higher in women with a recent diagnosis of menorrhagia. Patient education and counseling regarding the potential expulsion risk is recommended at insertion. The absolute risk of perforation for women with a recent diagnosis of menorrhagia is very low. The increased expulsion and perforation rates observed are likely because of causal factors of menorrhagia.

Development and assessment of a natural language processing model to identify residential instability in electronic health records' unstructured data: a comparison of 3 integrated healthcare delivery systems.

OBJECTIVE: To evaluate whether a natural language processing (NLP) algorithm could be adapted to extract, with acceptable validity, markers of residential instability (ie, homelessness and housing insecurity) from electronic health records (EHRs) of 3 healthcare systems. MATERIALS AND METHODS: We included patients 18 years and older who received care at 1 of 3 healthcare systems from 2016 through 2020 and had at least 1 free-text note in the EHR during this period. We conducted the study independently; the NLP algorithm logic and method of validity assessment were identical across sites. The approach to the development of the gold standard for assessment of validity differed across sites. Using the EntityRuler module of spaCy 2.3 Python toolkit, we created a rule-based NLP system made up of expert-developed patterns indicating residential instability at the lead site and enriched the NLP system using insight gained from its application at the other 2 sites. We adapted the algorithm at each site then validated the algorithm using a split-sample approach. We assessed the performance of the algorithm by measures of positive predictive value (precision), sensitivity (recall), and specificity. RESULTS: The NLP algorithm performed with moderate precision (0.45, 0.73, and 1.0) at 3 sites. The sensitivity and specificity of the NLP algorithm varied across 3 sites (sensitivity: 0.68, 0.85, and 0.96; specificity: 0.69, 0.89, and 1.0). DISCUSSION: The performance of this NLP algorithm to identify residential instability in 3 different healthcare systems suggests the algorithm is generally valid and applicable in other healthcare systems with similar EHRs. CONCLUSION: The NLP approach developed in this project is adaptable and can be modified to extract types of social needs other than residential instability from EHRs across different healthcare systems.

An automated algorithm using free-text clinical notes to improve identification of transgender people.

Accurate identification of transgender persons is a critical first step in conducting transgender health studies. To develop an automated algorithm for identifying transgender individuals from electronic medical records (EMR) using free-text clinical notes. The development and validation of the algorithm was based on data from an integrated healthcare system that served as a participating site in the multicenter Study of Transition Outcomes and Gender. The training and test datasets each contained a total of 300 individuals identified between 2006 and 2014. Both datasets underwent a full medical record review by experienced research abstractors. The validated algorithm was then implemented to identify transgender individuals in the EMR using all clinical notes of patients that received care between January 1, 2015 and June 30, 2018. Validation of the algorithm against the full chart review demonstrated a high degree of accuracy with 97% sensitivity, 95% specificity, 94% positive predictive value, and 97% negative predictive value. The algorithm classified 7,409 individuals (3.5%) as "Definitely transgender" and 679 individuals (0.3%) as "Probably transgender" out of 212,138 candidates with a total of 378,641 clinical notes. The computerized NLP algorithm can support essential efforts to improve the health of transgender people.

Identifying Ectopic Pregnancy in a Large Integrated Health Care Delivery System: Algorithm Validation.

BACKGROUND: Surveillance of ectopic pregnancy (EP) using electronic databases is important. To our knowledge, no published study has assessed the validity of EP case ascertainment using electronic health records. OBJECTIVE: We aimed to assess the validity of an enhanced version of a previously validated algorithm, which used a combination of encounters with EP-related diagnostic/procedure codes and methotrexate injections. METHODS: Medical records of 500 women aged 15-44 years with membership at Kaiser Permanente Southern and Northern California between 2009 and 2018 and a potential EP were randomly selected for chart review, and true cases were identified. The enhanced algorithm included diagnostic/procedure codes from the International Classification of Diseases, Tenth Revision, used telephone appointment visits, and excluded cases with only abdominal EP diagnosis codes. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall performance (Youden index and F-score) of the algorithm were evaluated and compared to the validated algorithm. RESULTS: There were 334 true positive and 166 true negative EP cases with available records. True positive and true negative EP cases did not differ significantly according to maternal age, race/ethnicity, and smoking status. EP cases with only one encounter and non-tubal EPs were more likely to be misclassified. The sensitivity, specificity, PPV, and NPV of the enhanced algorithm for EP were 97.6%, 84.9%, 92.9%, and 94.6%, respectively. The Youden index and F-score were 82.5% and 95.2%, respectively. The sensitivity and NPV were lower for the previously published algorithm at 94.3% and 88.1%, respectively. The sensitivity of surgical procedure codes from electronic chart abstraction to correctly identify surgical management was 91.9%. The overall accuracy, defined as the percentage of EP cases with correct management (surgical, medical, and unclassified) identified by electronic chart abstraction, was 92.3%. CONCLUSIONS: The performance of the enhanced algorithm for EP case ascertainment in integrated health care databases is adequate to allow for use in future epidemiological studies. Use of this algorithm will likely result in better capture of true EP cases than the previously validated algorithm.

Characterization of patients with advanced chronic pancreatitis using natural language processing of radiology reports.

STUDY AIM: To develop and apply a natural language processing algorithm for characterization of patients diagnosed with chronic pancreatitis in a diverse integrated U.S. healthcare system. METHODS: Retrospective cohort study including patients initially diagnosed with chronic pancreatitis (CP) within a regional integrated healthcare system between January 1, 2006 and December 31, 2015. Imaging reports from these patients were extracted from the electronic medical record system and split into training, validation and implementation datasets. A natural language processing (NLP) algorithm was first developed through the training dataset to identify specific features (atrophy, calcification, pseudocyst, cyst and main duct dilatation) from free-text radiology reports. The validation dataset was applied to validate the performance by comparing against the manual chart review. The developed algorithm was then applied to the implementation dataset. We classified patients with calcification(s) or ≥2 radiographic features as advanced CP. We compared etiology, comorbid conditions, treatment parameters as well as survival between advanced CP and others diagnosed during the study period. RESULTS: 6,346 patients were diagnosed with CP during the study period with 58,085 radiology studies performed. For individual features, NLP yielded sensitivity from 88.7% to 95.3%, specificity from 98.2% to 100.0%. A total of 3,672 patients met cohort inclusion criteria: 1,330 (36.2%) had evidence of advanced CP. Patients with advanced CP had increased frequency of smoking (57.8% vs. 43.0%), diabetes (47.6% vs. 35.9%) and underweight body mass index (6.6% vs. 3.6%), all p<0.001. Mortality from pancreatic cancer was higher in advanced CP (15.3/1,000 person-year vs. 2.8/1,000, p<0.001). Underweight BMI (HR 1.6, 95% CL 1.2, 2.1), smoking (HR 1.4, 95% CL 1.1, 1.7) and diabetes (HR 1.4, 95% CL 1.2, 1.6) were independent risk factors for mortality. CONCLUSION: Patients with advanced CP experienced increased disease-related complications and pancreatic cancer-related mortality. Excess all-cause mortality was driven primarily by potentially modifiable risk factors including malnutrition, smoking and diabetes.

Antibiotic and antifungal treatment among persons with confirmed coccidioidomycosis - Southern California, 2011.

We investigated coccidioidomycosis testing and treatment patterns among persons in an integrated healthcare delivery system to identify gaps in diagnosis and treatment. Coccidioidomycosis diagnosis delays were common. Among persons who tested positive, 70% were prescribed antibiotics before positive coccidioidomycosis tests. Antibiotic treatment decreased and antifungal treatment increased after positive testing.

The Application of Community-Based Information from the American Community Survey in a Large Integrated Health Care Organization.

BACKGROUND: The American Community Survey (ACS) is the largest household survey conducted by the US Census Bureau. We sought to describe the community-level characteristics derived from the ACS among enrollees of Kaiser Permanente Southern California (KPSC), evaluate the associations between ACS estimates and selective individual-level health outcomes, and explore how using different scales of the census geography and the linearity assumption affect the associations. METHODS: We examined the associations between track-level and block group-level ACS 5-year estimates and 4 individual-level Healthcare Effectiveness Data and Information Set (HEDIS) outcome measures (comprehensive diabetes care, postpartum care, antidepressant medication management, and childhood immunization status) using multilevel generalized linear models. Odds ratios and their 95% confidence intervals were estimated for every 10% increase in ACS measures. RESULTS: 6,357,841 addresses were successfully geocoded to at least the tract level. The community-level demographic, socioeconomic, residential, and other ACS measures varied among KPSC health plan enrollees. A majority of these ACS measures were associated with the selected HEDIS health outcomes. The directions of the effects were consistent across health outcomes; however, the magnitudes of the effect sizes varied. Within each HEDIS health outcome, the relative size of the effects appeared to remain similar. Differences between the census tract- and block group-level estimates were minor, especially for measures related to race/ethnicity, education, income, and occupation. CONCLUSION: These findings support the use of many ACS measures at neighborhood levels to predict health outcomes. The geographic units might have little effect on the results. The linearity assumption should be made with caution.

Temporal Trends in Mortality Rates among Kaiser Permanente Southern California Health Plan Enrollees, 2001-2016.

BACKGROUND: Temporal analyses of death rates in the US have found a decreasing trend in all-cause and major cause-specific mortality rates. OBJECTIVES: To determine mortality trends in Kaiser Permanente Southern California (KPSC), a large insured population, and whether they differ from those of California and the US. METHODS: Trends in age-adjusted all-cause and cause-specific mortality rates from 2001 to 2016 were determined using data collected in KPSC and those derived through linkage with California State death files and were compared with trends in the US and California. Trends of race/ethnicity-specific all-cause and cause-specific mortality rates were also examined. Average annual percent changes (AAPC) and 95% confidence intervals (CI) were calculated. RESULTS: From 2001 to 2016, the age-adjusted all-cause mortality rate per 100,000 person-years decreased significantly in KPSC (AAPC = -1.84, 95% CI = -2.95 to -0.71), California (AAPC = -1.60, 95% CI = -2.51 to -0.69) and the US (AAPC = -1.10, 95% CI = -1.78 to -0.42). Rates of 2 major causes of death, cancer and heart disease, also decreased significantly in the 3 populations. Differences in trends of age-adjusted all-cause mortality rates and the top 10 cause-specific mortality rates between KPSC and California or the US were not statistically significant at the 95% level. No significant difference was found in the trends of race/ethnicity-specific, sex-specific, or race/ethnicity- and sex-specific all-cause mortality rates between KPSC and California or the US. CONCLUSION: Trends in age-adjusted mortality rates in this insured population were comparable to those of the US and California.