Abstract
Introduction
The evidence on complication rates after gynecological surgery is based on multiple types of studies, and the level of evidence is generally low. We aimed to validate the registration of complications in the Swedish National Quality Register of Gynecological Surgery (GynOp), by cross‐linkage to multiple national registers.
Material and Methods
A national register‐based study using prospectively collected data was conducted, including women who had surgery on the uterus or adnexa for benign indications from January 1, 2017, to December 31, 2020. In Sweden, complications after gynecological surgery are registered in GynOp, and if the complication has rendered any interaction with healthcare, it is also in national health registers. The GynOp register, the National Patient Register, the Prescribed Drugs Register, and the Cause of Death Register were cross‐linked. Complications in GynOp and complications according to ICD10 were analyzed, as well as the cause of death if occurring within 3 months of surgery and prescription of antibiotics ≤30 days. Comparisons between the registries were descriptive.
Results
During the study period 32 537 surgeries were performed, of which 26 214 (80.6%) were minimally invasive. Complications were reported in GynOp for 569 women (1.7%) at surgery, 1045 (3.2%) while admitted, and 3868 (13.7%) from discharge to 3 months after surgery. In comparison, according to the Patient Register, 2254 women (6.9%) had postoperative complications within 3 months of discharge (difference of 6.8 percentage points [95% confidence interval 6.2–7.2]). Furthermore, 4117 individuals (12.7%) had a prescription of antibiotics ≤30 days which could indicate a postoperative infection. The rates of hemorrhage, wound dehiscence, and thrombosis were comparable between GynOp and the Patient Register while diagnoses not leading to contact with specialized care had higher rates in the quality register. The coverage of complications was 79.1% in GynOp and 46.1% in the Patient Register, using the total number of complications from both registers as the denominator.
Conclusions
A higher frequency of complications is captured in GynOp than in the National Patient Register. Patient‐reported outcomes assessed by a physician are beneficial in identifying complications indicating the importance of structured pre‐defined follow‐up over a set period.
Keywords: complications, gynecological surgery, patient register, patient‐reported outcome, quality register, validation
GynOp, a national quality register for gynecological surgery captures more complications than the national health registers. A structured follow‐up, including patient reported and physician assessed complications after surgery, is beneficial in describing the full spectrum of postoperative complications.
Abbreviations
- CD
Clavien Dindo classification
- DORS
the Cause of Death Register
- GynOp
The National Quality Register of Gynecological Surgery
- ICD10
International Classification of Diseases, Tenth Revision
- LMED
the Swedish Prescribed Drug Register
- PAR
the National Patient Register
- SSI
surgical site infection
- UTI
urinary tract infection
Key message.
GynOp, a national quality register for gynecological surgery captures more complications than the national health registers. A structured follow‐up, including patient‐reported and physician‐assessed complications after surgery, is beneficial in describing the full spectrum of postoperative complications.
1. INTRODUCTION
Any surgery is associated with a risk of complications, and it has long been acknowledged that the reporting of complications is inadequate. 1 , 2 There is no consistency in how the data is collected, in what time period after surgery a complication should be registered, and even what is regarded as a complication. 1 , 3 Complication rates from observational studies are based on individual hospital records, quality registers, insurance systems, and national statistics databases with varying coverage. 4 , 5 The methods for gathering data on the outcome of surgery differ, from using random sampling to aiming to register all cases. 5 , 6 Furthermore, complications are sometimes registered as composite outcomes, i.e. a collection of diagnoses gathered under an overarching name, which reduces comparability. 7 Hence, observational studies and systematic reviews based on these studies are only as trustworthy as the input to these registers. 8 In comparison, randomized controlled trials (RCT) are commonly declared to be the gold standard for comparing interventions. However, many trials are designed to evaluate treatment efficacy or effectiveness and are not powered to provide data on complications, in particular rare events. 8 , 9
The reported frequencies of complications after gynecological surgery vary, partly due to the inherent weaknesses in distinct types of studies. As an example, the rate of minor/major complications after laparoscopic hysterectomies is reported in an RCT from the UK to be 25.2%/11.1%, 10 vs 12.1%/4.1% in a register‐based study from Denmark 11 There are two large meta‐analyses by Sandberg et al. 12 and Pickett et al., 9 that review the outcomes of different surgical methods of hysterectomy. However, there is no information on observation time in the included publications, and the rationale for the choice of studied complications is lacking in the reviews. The evidence regarding complications after hysterectomy is rated as low. 9 , 12
Individuals residing in Sweden are allocated a unique personal identification number at birth or immigration. 13 The identification number is used in all national databases and in healthcare and provides the possibility to link data from multiple sources. 14 According to Swedish law, all contacts with healthcare must be documented and the recorded data is used in healthcare as well as in national health registers. 15 Data from healthcare can also be registered in specified quality registers depending on the diagnosis or performed intervention, and all individuals are included unless actively declining participation. The National Quality Register of Gynecological Surgery (GynOp), 15 , 16 is the main source for statistics about and outcomes after gynecological surgery in Sweden. However, the register is insufficiently validated. 17
The main aim of this study was to validate the registration of complications in GynOp by comparing the frequency and type of complications after benign gynecological surgery registered in GynOp with the National Patient Register (PAR), the Cause of Death Register (DORS) and the Swedish Prescribed Drug Register (LMED).
2. MATERIAL AND METHODS
We conducted a national register‐based study using prospectively collected data from four national registers, including women who had abdominal, laparoscopic, robot‐assisted laparoscopic, or vaginal surgery of the uterus or adnexa, with benign indications from January 1, 2017, to December 31, 2020. The date of surgery was collected from GynOp, and data from GynOp was crosslinked with PAR, LMED, and DORS. Specified variables were collected from each register, up to 3 months after surgery (Figure 1). Data were pseudonymized, with the code key not available to the researchers.
FIGURE 1.
Flowchart of selection of data from the Swedish National Quality Register of Gynecological Surgery, the Swedish National Patient Register, the Swedish Prescribed Drug Register, and the Swedish Cause of Death Register.
GynOp consists of six sub‐registers for different types of benign gynecological surgery; Hysterectomy, Adnexal surgery, Urinary incontinence, Pelvic organ prolapse, Intra‐uterine surgery, and Perineal Laceration. The register collects information regarding the date and type of surgery including intraoperative measurements, eg bleeding amount or presence of adhesions. Moreover, indications, demographic and health characteristics of the women, patient satisfaction, and the presence and classification of any complications up to 1 year after surgery are registered. 18 The coverage for all Swedish national quality registers is calculated by linkage with PAR using the Swedish personal identification number, and GynOp's coverage of all gynecological surgeries in Sweden was 88.6% in 2019. 6 The GynOp register is certified at the highest level by the organization Swedish National Quality Registers, which is supported by the Government and the Swedish Association of Local Authorities and Regions. Furthermore, it is approved by the Swedish data protection agency. 16 In GynOp, complications are registered by a physician based on type, severity, the Clavien‐Dindo classification, 17 , 18 and timepoint (at surgery, at discharge from hospital, at 3 months and at 1 year after surgery). The patients report subjective assessments of surgical outcomes and adverse events to the register through web‐based questionnaires after two, up to three, months and at 1 year after surgery. The physician's postoperative assessment at 3 months and 1 year can be based on a combination of sources; the patient questionnaire, phone contact, visit to the clinic, and healthcare charts. The complications from GynOp presented in this study are physician assessed complications.
The recommended definitions for severity grading by GynOp are presented in Table S1, and the type of complication in Table S2. Any surgical intervention to treat a complication can be reported in GynOp via the original method, i.e. a check box “Complication has been treated by new surgery” followed by a scroll list with a selection of possible surgical interventions, and a Clavien‐Dindo classification (CD) of 3A or 3B. GynOp collects data on the date of death in an automated fashion but the cause of death, and if it is to be regarded as a complication, is registered through the physician's entry to the register. Data was collected from GynOp up to 3 months after surgery to cover the period when the patient questionnaire, including reminders, was sent and could be assessed.
PAR has collected data on diagnoses and procedures from specialized healthcare contacts since 2001. In 2021, a main diagnosis was recorded in 99.2% and 97.6% of all registered in‐ and out‐patient contacts, respectively. Codes for performed medical procedures were reported in 65% of hospital admissions and 46.4% of outpatient contacts, 6 , 19 where a missing code can signify either no procedure performed or code not reported. Regarding gynecological surgeries in Sweden, the coverage for both PAR and GynOp is calculated via cross‐linkage of the two registers, where the denominator is the cases registered in any, or both, of the registers. PAR had a coverage of 85.5% of benign gynecological procedures performed in 2019. 6 In this study data from PAR provide information if specified ICD10 codes (postoperative infection, thrombosis, wound complications, etc.) were registered up to 3 months after surgery (Table S3).
LMED contains information on all prescribed pharmaceuticals dispensed in pharmacies since 2005 20 and has a coverage of >99%. 21 In linkage with the data from GynOp, LMED provided information if a prescription of specified antibiotics was dispensed ≤30 days of surgery. Prespecified ATC codes for antibiotics potentially used for postoperative complications such as urinary tract infections (UTI), pneumonia, sepsis, abscess, or surgical site infection (SSI) were collected (Table S3). In addition, a specified list of infections that are not regarded as postoperative complications was requested from PAR for exclusion of cases indicated by data from LMED, eg otitis media. (Figure 1, Table S3).
DORS contains the date of death for all diseased residents in Sweden, with a cause of death registered in >99%. 22 Date and cause of death (ICD10) for any deaths occurring up to 3 months after surgery was requested from DORS, to capture any mortal complications. The selected variables from the different databases are presented in Table S3.
2.1. Statistical analyses
Continuous variables are summarized using median and quartiles (Q1; Q3). Categorical variables are summarized as numbers and percentages. Complication frequencies <3 are not disclosed to protect privacy. Validation of complications registered in GynOp against a diagnosis of complications, according to ICD10 as reported in PAR, or a prescription of antibiotics, is descriptive. Coverage of any registered complication within 3 months after surgery in GynOp and/or PAR was calculated using physician‐registered complication in GynOp and ICD10 code (T81.0–9) of a postoperative complication from PAR. Confidence intervals for the difference in complication rates between GynOp and PAR, or LMED, were estimated using the Miettinen‐Nurminien method. 23
A subgroup analysis of potential differences between responders and non‐responders to the GynOp postoperative questionnaire was conducted. For comparison between groups, Student's t‐test was used for continuous variables, and the chi‐squared test for proportions in categorical variables. A two‐sided p value of <0.05 was selected to indicate statistical significance. SPSS (IBM Corp. IBM SPSS Statistics, version 28.0) was used.
3. RESULTS
A total of 31 974 individuals with 32 537 surgeries from 1 January 2017 to 31 December 2020 were included. The age range was 18–96 years, with a median age of 44 years. A new surgical procedure was registered in GynOp for 56 patients within the observed 3 months after surgery, most of which were not registered as complications due to the prior surgery. Minimally invasive surgery (laparoscopic, robot‐assisted laparoscopic, or vaginal) was performed in 26 214 cases (80.6%), and 17 047 cases (52.4%) were registered as adnexal surgery only. At the time of surgery, 31 627 (97.2%) of the patients were assessed as ASA 1–2 (American Society of Anesthesiologists) (Table 1).
TABLE 1.
Background characteristics of women with surgery of uterus or adnexa with benign indications, 2017–2020 in Sweden (N = 32 537).
| Median | Q1; Q3 | Min–Max | |
|---|---|---|---|
| Age | 44.0 | 35.0–52.0 | 18–96 |
| Body Mass Index a | 25.3 | 22.6–28.9 | 15.0–69.9 |
| N | % | |
|---|---|---|
| ASA | ||
| ASA 1 | 21 002 | 64.5 |
| ASA 2 | 10 625 | 32.7 |
| ASA 3 | 876 | 2.7 |
| ASA 4–5 | 24 | 0.1 |
| Smoking a | ||
| No smoking | 24 462 | 89.1 |
| Smoking | 2988 | 10.9 |
| Surgical procedures (N) b | ||
| 1 | 31 974 | 98.3 |
| 2 | 541 | 1.7 |
| 3–4 | 22 | 0.1 |
| Type of surgery | ||
| Adnex | 12 060 | 37.1 |
| Uterine | 9504 | 29.2 |
| Adnex and uterine | 8833 | 27.1 |
| Prolapse | 2120 | 6.5 |
| Urinary incontinence | 20 | 0.1 |
| Surgical approach | ||
| Laparotomy | 6323 | 19.4 |
| Laparoscopy | 19 746 | 60.7 |
| Robot‐assisted laparoscopy | 3162 | 9.7 |
| Vaginal surgery | 3306 | 10.2 |
| Year of surgery | ||
| 2017 | 7335 | 22.5 |
| 2018 | 8502 | 26.1 |
| 2019 | 8995 | 27.6 |
| 2020 | 7705 | 23.7 |
Variables missing for body mass index in 16.3% and smoking in 15.6% of cases. Percentage is calculated on valid data.
Number of surgeries for each patient within the study period 2017–2020.
3.1. Completeness and overall complication rates
The completeness in reporting if a complication had occurred was 100% perioperatively, 99.5% at discharge, and 87.0% at assessment within 3 months in data from GynOp. The physician assessments at 3 months were based on information from the questionnaire in 91.4% of cases, from medical records in 18.5%, and from personal contact in 3.4% (a combination of sources was possible).
The overall number of complications according to GynOp was 569 (1.7%) perioperatively, 1045 (3.2%) during the inpatient stay, and 3868 (13.7%) from discharge to the assessment at approximately 3 months (Table 2). According to PAR, complications (a specified main diagnosis of a postoperative complication (ICD10 T81.0–T81.9) or organ damage (bladder, ureter, or urethra)) occurred in 2054 cases (6.3%) within the first month after discharge and in 2254 cases (6.9%) up to 3 months postop. When comparing complications from discharge up to 3 months between GynOp and PAR, the captured rate in GynOp was 6.8 percentage points (pp) higher (95% CI 6.2–7.2; Table 3). The coverage of reported complications from discharge to 3 months after surgery was 79.1% (3868/4888) in GynOp, and 46.1% (2254/4888) in PAR, using the total number of complications from both registers as the denominator. (Figure 2). Only 25.2% (1234/4888) of all complications were represented in both registers. When dividing the registered complications in GynOp by severity, 61.0% (322/528) of individuals with a major complication in GynOp had a complication registered in PAR, in comparison with 27.5% (930/3380) of those with a registered minor complication.
TABLE 2.
Complications from the Swedish National Quality Register of Gynecological Surgery, the National Patient Register, the Prescribed Drugs Register and the Cause of Death Register, from surgery to 3 months postop.
| 32 537 Surgeries | The Swedish National Quality Register of Gynecological Surgery—GynOp a | The Swedish National Registries—patient, prescribed drugs and cause of death registries | |||
|---|---|---|---|---|---|
| Peroperatively, N (%) | During hospital stay, N (%) | From discharge to 3 months, N (%) | From discharge to 1 month, N (%) | From discharge to 3 months, N (%) | |
| Complication | 569 (1.7) | 1045 (3.2) | 3868 (13.7) | 2054 b (6.3) | 2254 (6.9) |
| Thrombosis c | 15 (0) | 30 (0.1) | 49 (0.2) | 79 (0.2) | |
| Wound dehiscence | 12 (0) | 67 (0.2) | 88 (0.3) | 115 (0.4) | |
| Hemorrhage | 228 (0.7) | 338 d (1.0) | 683 (2.4) | 687 (2.1) | 750 (2.3) |
| Organ damage e | 371 (1.1) | 180 (0.6) | 591 (2.1) | 68 f (0.2) | 82 (0.3) |
| Cardiac disease g | 17 (0) | 4 (0) | <3 (0) | 6 (0) | |
| CNS disease g | 10 (0) | 7 (0) | 4 (0) | 6 (0) | |
| Gastrointestinal problems g | 43 (0.1) | 167 (0.6) | 48 (0.1) | 60 (0.2) | |
| Urinary retention | 153 (0.4) | 94 (0.3) | 86 (0.3) | 97 (0.3) | |
| Urinary incontinence | <3 (0) | 198 (0.7) | 7 (0) | 20 (0.1) | |
| Allergic reaction h | 7 (0) | 3 (0) | |||
| Postoperative pain h | 83 (0.2) | 923 (3.3) | |||
| Infection (treated by antibiotics) | 142 (0.5) | 2414 (8.5) | 4117 i (12.7) | ||
| Urinary tract infection | 27 (0.1) | 831 (2.9) | 64 (0.2) | 97 (0.3) | |
| Pneumonia | 8 (0) | 3 (0) | 3 (0) | 11 (0) | |
| Surgical site infection | 63 (0.2) | 1005 (3.5) | 1110 (3.4) | 1223 (3.8) | |
| Death | 0 (0) | 0 (0) | <3 (0) | 8 (0) | |
| New surgery to treat complication | 144 (0.4) | 117 (0.4) | 219 (0.7) | 229 (0.7) | |
| Wound dehiscence | 5 (0) | 26 (0.1) | 30 (0.1) | 36 (0.1) | |
| Infection | 9 (0) | 7 (0) | 28 (0.1) | 29 (0.1) | |
| Hemorrhage | 90 (0.3) | 17 (0) | 144 (0.4) | 146 (0.4) | |
| Other | 40 (0.1) | 67 (0.2) | 23 (0.1) | 24 (0.1) | |
| Clavien Dindo | N = 1041 | N = 2711 |
|---|---|---|
| 370 (1.1) | 1253 (4.4) | |
| 2 | 316 (1.0) | 1177 (4.2) |
| 3A | 137 (0.1) | 115 (0.4) |
| 3B | 210 (0.6) | 164 (0.6) |
| 4A | 5 (0) | <3 (0) |
| 4B | 3 (0) | 0 (0) |
Cases with assessment of complications; 100% peroperative (N = 32 537), 99.5% at discharge (N = 32 370), 87.0% (N = 28 319) at 3 months. Complications from GynOp are only presented once in the table, ie data is not cumulative.
Complication in the Patient Register is the use of the specific ICD10 codes for postoperative complication (T81.1‐9) alt. ICD10 code for bladder, ureter or urethra damage (S371.1‐3).
Pulmonary embolus or deep vein thrombosis.
Postoperative hemorrhage in GynOp is divided into vaginal, abdominal wall, intraabdominal and anemia, further detail in Table 5 and Table S2.
Organ damage according to GynOp is divided into vagina, uterus/adnexa, urethra, bladder, ureter, intestines, blood vessel, and nerve, further detail in Table 5 and Table S2.
Organ damage from the Patient Register is ICD10 T81.2 (postoperative organ damage) or S371.1‐3 (bladder, urethra or ureter damage).
Cardiac disease, CNS disease and Gastrointestinal disease are complications in GynOp, in the Patient Register Cardiac disease represents coronary infarction, CNS disease represents stroke and gastrointestinal disease represents ileus.
The ICD10 codes for pain and allergic reactions were not requested from the Patient Register, no comparison possible.
According to the Prescribed Drugs Register, further details in Table 4.
TABLE 3.
Comparison of complication rates between the Swedish National Quality Register of Gynecological Surgery and the National Patient Register, from discharge to 3 months postop. Gynecological surgeries in Sweden 2017–2020 (N = 32 537).
| The Swedish National Quality Register of Gynecological Surgery—GynOp, N (%) | The Swedish National Registries—patient, prescribed drugs and cause of death registries, N (%) | Difference in complication rates (percentage points) with 95% confidence interval | |
|---|---|---|---|
| Complication | 3868 (13.7) | 2254 (6.9) | 6.8 [6.2 to 7.2] |
| Thrombosis b | 30 (0.1) | 79 (0.2) | −0.1 [−0.2 to −0.1] |
| Wound dehiscence | 67 (0.2) | 115 (0.4) | −0.2 [−0.2 to 0.0] |
| Hemorrhage c | 683 (2.4) | 750 (2.3) | 0.1 [−0.1 to 0.3] |
| Organ damage d , e | 591 (2.1) | 82 (0.3) | 1.8 [1.7 to 2.0] |
| Cardiac disease f | 4 (0.0) | 6 (0.0) | a |
| CNS disease f | 7 (0.0) | 6 (0.0) | a |
| Gastrointestinal problems f | 167 (0.6) | 60 (0.2) | 0.4 [0.3 to 0.5] |
| Urinary retention | 94 (0.3) | 97 (0.3) | 0 [−0.1 to 0.1] |
| Urinary incontinence | 198 (0.7) | 20 (0.1) | 0.6 [0.5 to 0.7] |
| Infection (treated by antibiotics) | 2414 (8.5) | ||
| Urinary tract infection | 831 (2.9) | 97 (0.3) | 2.6 [2.4 to 2.8] |
| Pneumonia | 3 (0.0) | 11 (0.0) | a |
| Surgical site infection | 1005 (3.5) | 1223 (3.8) | −0.3 [−0.5 to 0.1] |
| Death | < 3 | 8 (0.0) | a |
| New surgery to treat complication | 117 (0.4) | 229 (0.7) | −0.3 [−0.4 to −0.2] |
| Wound dehiscence | 26 (0.1) | 36 (0.1) | 0 [−0.1 to 0.0] |
| Infection | 7 (0.0) | 29 (0.1) | a |
| Hemorrhage | 17 (0.0) | 146 (0.4) | −0.4 [−0.5 to −0.3] |
| Other | 67 (0.2) | 24 (0.1) | 0.1 [0.1 to 0.2] |
Note: Complications in GynOp are based on physicians' assessment and registration in the register. Complication in the Patient Register is the use of the specific ICD10 codes for postoperative complication (T81.1–9) alt. ICD10 code for bladder, ureter or urethra damage (S371.1–3).
Confidence interval not estimated due to low number of cases.
Pulmonary embolus or deep vein thrombosis.
Postoperative hemorrhage in GynOp is divided into vaginal, abdominal wall, intraabdominal and anemia, further details in Table 5 and Table S2.
Organ damage according to GynOp is divided into vagina, uterus/adnexa, urethra, bladder, ureter, intestines, blood vessel, and nerve, further details in Table 5 and Table S2.
Organ damage from the Patient Register is ICD10 T81.2 (postoperative organ damage) or S371.1–3 (bladder, urethra or ureter damage).
Cardiac disease, CNS disease and Gastrointestinal disease are complications in GynOp, in the Patient Register Cardiac disease represents coronary infarction, CNS disease represents stroke and gastrointestinal disease represents ileus.
FIGURE 2.
Completeness of registered complications in the Swedish National Quality Register of Gynecological Surgery (GynOp) and the Swedish National Patient Register (PAR). Captured complications within 3 months of benign gynecological surgery in Sweden 2017–2020, in the two registers.
3.2. Specific complications
The most common complication according to GynOp was SSI (3.5%), followed by postoperative pain (3.3%), UTI (2.9%), and hemorrhage (2.4%). There was no significant difference in the rates of SSI (−0.3 pp [95% CI −0.5 to 0.1]) or hemorrhage (0.1 pp [95% CI −0.1 to 0.3]) between GynOp and PAR (Table 3). No ICD10 code for pain was requested from PAR, hence no comparison of pain as a complication can be made between GynOp and PAR. According to LMED 4117 individuals (12.7%) had a dispensed prescription of antibiotics ≤30 days of surgery. The LMED does not indicate the prescription; however, the main indications for the specified antibiotics and the frequency of the dispensations can be seen in Table 4. Antibiotics used to solely treat UTIs correspond to approximately one‐third of prescriptions. UTIs were registered in GynOp in 2.9% of surgeries vs 0.3% in PAR (difference 2.6 pp [95% CI 2.4–2.8]) (Table 3).
TABLE 4.
Dispensed antibiotics ≤30 days of 32 537 benign uterine and adnexal procedures, according to the Prescribed Drugs Register. a
| Dispensed antibiotic | Main indications according to Swedish treatment recommendations | N | % of prescriptions | % of 32 537 surgeries |
|---|---|---|---|---|
| Nitrofurantoin | UTI (cystitis) | 491 | 11.9 | 1.5 |
| Pivmecillinam | UTI (cystitis) | 812 | 19.7 | 2.5 |
| Trimetoprim | UTI (cystitis) | 42 | 1.0 | 0.1 |
| Cefadroxil | UTI b /Pneumonia/Wound infection/Endometritis | 342 | 8.3 | 1.1 |
| Erytromycin | Pneumonia | 16 | 0.4 | 0 |
| Ciprofloxacin | Pyelonephritis/Intraabdominal infection | 473 | 11.5 | 1.5 |
| Moxifloxacin | Pneumonia/Wound infection/ Intraabdominal infection | 1 | 0 | 0 |
| Levofloxacin | UTI/Pyelonephritis/Atypical Pneumonia | 3 | 0.1 | 0 |
| Clindamycin | Pneumonia/Wound infection/Intraabdominal infection | 367 | 8.9 | 1.1 |
| Sulfa/ Trimetoprim | Pneumonia/Pyelonephritis/Wound infection/Intraabdominal infection | 149 | 3.6 | 0.5 |
| Penicillin V | Pneumonia/Wound infection | 155 | 3.8 | 0.5 |
| Flucloxacillin | Wound infection | 522 | 12.7 | 1.6 |
| Amoxicillin | UTI/Pyelonephritis/Pneumonia | 88 | 2.1 | 0.3 |
| Amoxicillin and enzyme inhibitor | UTI/Pyelonephritis/Pneumonia/Wound infection/Intraabdominal infection/Endometritis | 530 | 12.9 | 1.6 |
| Doxycycline | Pneumonia/Intraabdominal infection/Endometritis | 126 | 3.1 | 0.4 |
| 4117 | 100 | 12.7 |
Individuals with a diagnosis (ICD10) registered in the Patient Register not regarded as postop complication but possibly rendering a prescription of antibiotics are excluded (appendicitis, gastric ulcer eradication, diverticulitis, cholecystitis, media otitis, sinusitis, tonsillitis etc.).
In case of multiple possible indications the most likely indication is marked in bold.
The occurrence of wound dehiscence or new surgery due to wound dehiscence was similar in both GynOp and PAR (Table 3). Hemorrhage, the most common cause for new surgery according to both registers, was also similarly reported (Table 2). Hemorrhage registered in GynOp 3 months after surgery is sub‐divided into vaginal, abdominal wall, intraabdominal, and not otherwise specified, where each individual can be registered under more than one option (Table 5). When comparing the available methods for reporting surgical interventions to treat a complication the CD classification generates the highest frequency. A surgical intervention to treat a complication was reported in 1% of cases according to CD (CD 3A or 3B) within 3 months of the initial surgery. According to the original method for reporting and classifying surgical intervention for complications in GynOp, new surgery was performed in 0.4% vs 0.7% in PAR, (−0.3 pp [95% CI −0.4 to −0.2]; Table 3).
TABLE 5.
Type of hemorrhage and organ damage according to the Swedish National Quality Register of Gynecological Surgery, registered as peroperative, during hospital stay and from discharge to 3 months after surgery (N = 32 537 surgeries).
| Peroperative, N (%) | During hospital stay, N (%) | From discharge to 3 months, N (%) | ||||
|---|---|---|---|---|---|---|
| Minor | Major | Minor | Major | Minor | Major | |
| Hemorrhage | 190 (0.6) | 38 (0.1) | ||||
| Vaginal | 42 (0.1) | 11 (0) | 262 (0.9) | 43 (0.2) | ||
| Abdominal wall | 56 (0.2) | 22 (0) | 165 (0.6) | 19 (0.1) | ||
| Intraabdominal | 89 (0.3) | 112 (0.4) | 108 (0.4) | 38 (0.1) | ||
| Other, not specified | 15 (0) | 3 (0) | 75 (0.2) | 26 (0.1) | ||
| Type of organ damage | ||||||
| Vascular | 10 (0) | 3 (0) | <3 (0) | <3 (0) | 9 (0) | 7 (0) |
| Intestine | 68 (0.2) | 40 (0.1) | 17 (0.1) | 27 (0.1) | 85 (0.3) | 46 (0.1) |
| Uterus or adnexa | 37 (0.1) | 4 (0) | 10 (0) | 1 (0) | 25 (0.1) | 7 (0) |
| Bladder | 104 (0.3) | 36 (0.1) | 52 (0.2) | 34 (0.1) | 112 (0.4) | 28 (0.1) |
| Ureter | 8 (0) | 18 (0) | 0 (0) | 39 (0.1) | 12 (0) | 36 (0.1) |
| Urethra | 0 (0) | <3 (0) | 0 (0) | 0 (0) | 46 (0.2) | 13 (0) |
| Vagina | <3 (0) | <3 (0) | 51 (0.2) | 22 (0.1) | ||
| Nerve | 4 (0) | 0 (0) | 81 (0.3) | 19 (0.1) | ||
| Wound dehiscence | 5 (0) | 7 (0) | 40 (0.1) | 27 (0.1) | ||
| Fistula | 3 (0) | <3 (0) | ||||
Note: Cases with assessment of complications; 100% peroperative (N = 32 537), 99.5% at discharge (N = 32 370), 87.0% (N = 28 319) at 3 months. Numbers under 3 is presented as <3 to protect privacy.
Thromboembolic events (pulmonary embolus or deep vein thrombosis) were registered for 15 individuals while still in the hospital and 30 more cases from discharge until follow‐up in GynOp. This corresponds to the entry in PAR of 49 cases up to one month post‐op and 79 cases up to 3 months. Organ damage up to 3 months after surgery was one of the main discrepancies, with 591 (2.1%) cases registered in GynOp and 82 (0.3%) in PAR (1.8 pp [95% CI 1.7–2.0]; Table 3). The registration of ureter damage was similar in both GynOp and PAR (0.1%), while bladder damage was registered at a higher frequency in GynOp (Table 5). The frequency of new surgery for other reasons than infection, hemorrhage, and wound dehiscence according to GynOp and PAR was 0.2% and 0.1%, respectively (Table 3).
Out of the 32 537 surgeries, eight women were diseased within 3 months of surgery. None of these women responded to the postoperative survey and no assessment if a complication had occurred, potentially leading to death, was made for either of them (data not presented for confidentiality reasons).
3.3. Subanalysis
The response rate of the patient questionnaire sent at 2 months (including reminders) was 82.9%. We conducted a sub‐analysis of responders vs non‐responders of the GynOp questionnaire regarding outcomes in the health registers from surgery up to 3 months (Table 6). The non‐responders were younger and had to a higher degree had outpatient surgery, as well as adnexal surgery only vs hysterectomy. They also had a higher frequency of prescribed antibiotics ≤30 days of surgery and registered complications in PAR. The non‐responders also presented a slightly higher rate of a UTI diagnosis.
TABLE 6.
Background data and complications in the National Patient Register and the Prescribed Drugs Register up to 3 months after surgery for responders vs non responders of GynOp's patient questionnaire, i.e. sub analysis of possible missing data from GynOp.
| Responders, N = 26 970 (82.9%) | Non‐responders, N = 5567 (17.1%) | p | |||
|---|---|---|---|---|---|
| Mean | 95% CI | Mean | 95% CI | ||
| BMI | 26.2 | 26.1–26.3 | 26.2 | 26.0–26.3 | 0.77 |
| Age | 45.6 | 45.4–45.8 | 40.8 | 40.4–41.1 | <0.001 * |
| N | % | N | % | p | |
|---|---|---|---|---|---|
| ASA 3–4 | 160 | 2.9 | 740 | 2.7 | 0.48 |
| Outpatient surgery | 9048 | 33.7 | 2075 | 37.7 | <0.001 * |
| Adnexal surgery only | 13 583 | 53.7 | 3464 | 64.4 | <0.001 * |
| Complications | |||||
| Complication (ICD10 T81.0–9) | 1787 | 6.6 | 434 | 7.7 | 0.002 * |
| Organ damage | 65 | 0.2 | 17 | 0.3 | 0.38 |
| Thrombosis | 64 | 0.2 | 15 | 0.3 | 0.66 |
| Hemorrhage | 606 | 2.2 | 144 | 2.6 | 0.12 |
| Ileus | 52 | 0.2 | 8 | 0.1 | 0.44 |
| Incontinence | 15 | 0.1 | 5 | 0.1 | 0.35 |
| Urinary retention | 74 | 0.3 | 23 | 0.4 | 0.08 |
| Coronary heart disease | 4 | 0 | 2 | 0 | 0.29 |
| Stroke | 5 | 0 | 1 | 0 | 0.98 |
| UTI | 72 | 0.3 | 25 | 0.4 | 0.02 * |
| New surgery | 182 | 0.7 | 47 | 0.8 | 0.17 |
| Prescription of antibiotics ≤30 days | 3291 | 12.2 | 826 | 14.9 | <0.001 * |
p < 0.05 is marked in italics.
4. DISCUSSION
GynOp, the Swedish National Quality Register of Gynecological Surgery, has its strengths and weaknesses in complication registration, as indicated by our results. One of its key strengths is the comprehensive collection of information at predefined intervals from both patients and physicians, leading to significantly higher coverage of complications in comparison to PAR. In GynOp, most postoperative complications are diagnosed after discharge up to 3 months after surgery, and the rate is higher than in PAR. Complications not requiring a healthcare contact or managed in primary care are to a higher degree registered in GynOp, where data input is supported by patient‐reported adverse events on set time intervals after surgery.
The strength of this study is the combination of four national registries with high coverage, all including prospectively collected data during four consecutive years. However, it is important to note some limitations. Contacts with primary care are not included in PAR, meaning general complications that do not require specialized care might not be captured. In this study, we used ICD10 codes to exclude possible non‐complication indications for dispensed antibiotics. Since all ICD10 codes originate from specialized care, there is a possibility of an overestimation on the number of prescriptions that might relate to a complication. Additionally, the Clavien‐Dindo classification was introduced in the register in March 2017 and was not in use for the first 3 months of the study period. This change might have impacted on the consistency of complication reporting during that time.
The main difference between GynOp as a quality register and any health register is the professional assessment regarding causal associations between surgery and complication in the quality register. That is not the case in the health register, unless using the specific ICD10 codes for postoperative complications. Diagnostic codes in a health register in the period after surgery do not necessarily relate to the performed intervention, and the further in time from surgery an adverse event is registered the higher the risk of false associations, as stated by Ladha et al. 24 As an example, a thrombosis which is registered in GynOp is regarded as a complication while in PAR it is a registered diagnosis. Hence, the total of 79 cases up to 3 months after surgery in PAR would also include those which might not be regarded as a postoperative complication by the assessing surgeon. A UK data linkage study by Bouras et al. found increasing numbers of thromboembolic events up to 90 days post‐surgery, describing linking of data from different registers as useful for evaluating postoperative complications. 4 The American College of Surgeon's National Surgical Quality Improvement Program, NSQIP, uses sampling of surgeries, and records complications up to 30 days after surgery. 5 In a study using data from NSQIP only 33.4% of thromboses occurred post‐discharge 25 in comparison to 65% by Bouras et al. This indicates that longer follow‐up is beneficial for identifying complications that are not distinguishable early postop eg pain, or have not developed yet, eg fistula or thrombosis. However, an assessment of the causal association of the procedure and the adverse event is necessary. Qureshi et al. conclude that reliable evidence of harm requires a synthesis of data from different sources, and the reporting of harm must be precise in date of onset, duration, and severity, 8 supporting this conclusion.
Another difference between a health register and GynOp as a quality register is the use of patient‐reported adverse events. Our results reveal comparable numbers between GynOp and PAR regarding SSI, while the diagnosis of a UTI is underestimated in PAR in comparison to the quality register. This can be explained by the fact that UTIs are mainly managed in primary care and thereby not registered in PAR but reported to GynOp as an adverse event in the follow‐up questionnaire. As for gastrointestinal and urinary problems, the rating as a complication in GynOp is much higher than the number of diagnoses in PAR. Complications that are registered as major in GynOp are in 61% represented in PAR, in comparison to minor complications which are present in 27.5% in PAR. Issues that do not prompt individuals to seek healthcare are more likely to be registered if patient reporting is used as a part of the routine follow‐up. 26 , 27 These types of adverse events are still important since they might affect both quality of life and satisfaction with the performed surgery. 27 The frequency of postoperative pain being regarded as a complication was 3.3% in GynOp and even higher than the number of UTI. In a study by Woodfield et al., postoperative pain was mainly captured by postoperative survey or phone contact and only rarely by hospital records. Yet, pain is by the patients perceived as a major complication, 27 and the development of chronic pain after a hysterectomy has been previously described. 28
It is, however, important to acknowledge possible bias in using patient reporting for outcomes. Using patient reporting of adverse events might give an overestimation if directly interpreted as complications, as these reported outcomes could be a natural consequence of the surgery. However, the method in GynOp is patient reporting as a basis for physician assessment on the presence of a complications, and this has been reported as a valuable method for increasing the capture of complications. 26 Moreover, our sub‐analysis revealed that non‐responders of the follow‐up survey were, on average, younger by approximately 5 years and had a higher incidence of outpatient surgery. Additionally, this group exhibited a higher frequency of prescribed antibiotics ≤30 days of surgery and more registered postoperative complications according to ICD10. These findings suggest that an incurred complication may reduce the motivation for filling out a postoperative questionnaire if there already has been a follow‐up contact due to the complication. This infers an underestimation of actual complications in GynOp, due to the non‐responders, even if the frequency of complications in the quality register is higher than the health register. To address this issue, patient reporting must be complemented by other methods to catch the outcome of non‐responders with potential complications. This is also supported by the finding that none of the individuals with a fatality within 3 months of their surgery, despite death being registered and regardless of the cause of death, was assessed for the presence of a complication in the quality register.
Both GynOp and PAR register if surgery has been required to treat a complication. In GynOp this is accomplished by two methods, the original, which is described in the methods section, and by Clavien‐Dindo. It appears that the CD classification for a complication needing surgical treatment is comparable with the data from PAR, but less so the original method from GynOp. It is possible that the addition of CD in March 2017 rendered clinicians less motivated to also use the original method, as the simplicity of use is essential in any follow‐up method.
A common method for reporting adverse events after surgery is the use of composite outcomes, i.e. gathering different diagnoses under an overarching title such as cardiac disease or organ damage. 7 , 24 Composite outcomes can be beneficial for an overall summary of effect, potentially enhancing statistical power and overall comprehension. 7 However, poorly defined they can lead to imprecision and be misleading. The discrepancy between GynOp and PAR regarding the diagnosis of organ damage in this study might be due to the imprecision of the diagnosis in both registers. As indicated by the editorial by Myles et al. there is a call for definitions of composite outcomes as well as core outcome sets to be reported in both research and clinical follow‐up of each type of surgery. 7 , 9 In a previous study, we found high interrater reliability when Swedish gynecologists rated clinical cases according to the methods in GynOp. However, the study also concluded that it is important to define what symptoms should be considered inherent to each type of surgery and which should be classified as complications. 17 The COMET initiative has an online database with relevant studies aimed at developing core outcome sets with several ongoing studies in gynecology, including hysterectomy. 29 Improving comparability of surgical outcomes, including complications, is essential in both research and clinical progress.
5. CONCLUSION
A higher frequency of complications is captured in GynOp than in the National Patient Register, most noticeable for complications that are not managed in specialized care. Combining GynOp data with national health registers can somewhat increase the number of captured complications. However, patient‐reported outcomes in combination with physicians' assessments structured as in GynOp are beneficial in identifying complications. This suggests that a structured predefined follow‐up over a set period is essential to enhance capture of complications after surgery.
AUTHOR CONTRIBUTIONS
Annika Idahl, Elin Collins, Per Liv, and Annika Strandell conceptualized the study, and the study protocol was revised and approved by all authors (Elin Collins, Annika Idahl, Annika Strandell, Per Liv, Maria Furberg, Sophia Ehrström and Mathias Pålsson). Data curation was performed by Per Liv and Elin Collins. Analysis was performed by Elin Collins and Per Liv. Elin Collins and Annika Idahl drafted the original manuscript, which was then revised and approved by all authors.
FUNDING INFORMATION
This work was funded by the County Council of Västerbotten (RV‐995813, RV‐981866, RV‐7000003), Lion's Cancer Research Foundation in Northern Sweden (LP‐22‐2314), and the Swedish Cancer Society (211408 Pj).
CONFLICT OF INTEREST STATEMENT
All authors report no conflict of interest.
ETHICS STATEMENT
Ethical approval was obtained from the Swedish ethical review authority on November 1, 2021 (Dnr. 2021‐05313‐01).
Supporting information
Table S1.
Table S2.
Table S3.
ACKNOWLEDGMENTS
We thank The Swedish Network for national clinical studies in Obstetrics and Gynecology (SNAKS) for their support. We acknowledge the Swedish National Quality Register of Gynecological Surgery (GynOp), The Swedish Government and the Swedish Association of Local Authorities and Regions, which both support GynOp.
Collins E, Liv P, Strandell A, et al. Complications after benign gynecologic surgery—How are they captured in register‐based research? A national register study in Sweden. Acta Obstet Gynecol Scand. 2024;103:2221‐2231. doi: 10.1111/aogs.14960
REFERENCES
- 1. Slankamenac K, Graf R, Barkun J, Puhan MA, Clavien PA. The comprehensive complication index: a novel continuous scale to measure surgical morbidity. Ann Surg. 2013;258:1‐7. [DOI] [PubMed] [Google Scholar]
- 2. Martin RC II, Brennan MF, Jaques DP. Quality of complication reporting in the surgical literature. Ann Surg. 2002;235:803‐813. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Sokol DK, Wilson J. What is a surgical complication? World J Surg. 2008;32:942‐944. [DOI] [PubMed] [Google Scholar]
- 4. Bouras G, Burns EM, Howell AM, Bottle A, Athanasiou T, Darzi A. Risk of post‐discharge venous thromboembolism and associated mortality in general surgery: a population‐based cohort study using linked hospital and primary care data in England. PLoS One. 2015;10:e0145759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Habermann EB, Harris AHS, Giori NJ. Large surgical databases with direct data abstraction: VASQIP and ACS‐NSQIP. J Bone Joint Surg Am. 2022;104:9‐14. [DOI] [PubMed] [Google Scholar]
- 6. Socialstyrelsen Bilaga Det statistiska registrets framställning och kvalitet‐ Patientregistret. Accessed 10 January 2024. https://www.socialstyrelsen.se/globalassets/sharepoint‐dokument/artikelkatalog/statistik/2022‐2‐7767‐tabeller.xlsx
- 7. Myles PS, Devereaux PJ. Pros and cons of composite endpoints in anesthesia trials. Anesthesiology. 2010;113:776‐778. [DOI] [PubMed] [Google Scholar]
- 8. Qureshi R, Mayo‐Wilson E, Li T. Harms in systematic reviews paper 1: an introduction to research on harms. J Clin Epidemiol. 2022;143:186‐196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Pickett CM, Seeratan DD, Mol BWJ, et al. Surgical approach to hysterectomy for benign gynaecological disease. Cochrane Database Syst Rev. 2023;8(8):CD003677. doi: 10.1002/14651858 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Garry R, Fountain J, Brown J, et al. EVALUATE hysterectomy trial: a multicentre randomised trial comparing abdominal, vaginal and laparoscopic methods of hysterectomy. Health Technol Assess. 2004;8:1‐154. [DOI] [PubMed] [Google Scholar]
- 11. Settnes A, Moeller C, Topsoee MF, et al. Complications after benign hysterectomy, according to procedure: a population‐based prospective cohort study from the Danish hysterectomy database, 2004‐2015. BJOG. 2020;127:1269‐1279. [DOI] [PubMed] [Google Scholar]
- 12. Sandberg EM, Twijnstra ARH, Driessen SRC, Jansen FW. Total laparoscopic hysterectomy versus vaginal hysterectomy: a systematic review and meta‐analysis. J Minim Invasive Gynecol. 2017;24:206‐217.e22. [DOI] [PubMed] [Google Scholar]
- 13. Ludvigsson JF, Otterblad‐Olausson P, Pettersson BU, Ekbom A. The Swedish personal identity number: possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol. 2009;24:659‐667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. The National Board of Health and Welfare. Health Data Registries. Accessed 14 November 2023. https://www.socialstyrelsen.se/statistik‐och‐data/register/
- 15. Socialdepartementet. Patientdatalagen. 7 kap 1§. https://www.riksdagen.se/sv/dokument‐lagar/dokument/svensk‐forfattningssamling/patientdatalag‐2008355_sfs‐2008‐355
- 16. Sveriges Kommuner och Regioner‐SKR. National Quality Registers. Accessed 10 January 2024. https://skr.se/en/kvalitetsregister/hittaregister/registerarkiv/gynekologiskaoperationer.44199.html
- 17. Collins E, Liv P, Strandell A, et al. Physicians' assessment of complications after gynecological surgery in Sweden: the GYNCOM survey. Acta Obstet Gynecol Scand. 2023;102:1479‐1487. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. GynOp. The National Quality Register of Gynecological Surgery. Accessed 10 January 2024. https://www.gynop.se/om‐gynop/
- 19. Socialstyrelsen. Det statistiska registrets framställning och kvalitet—Patientregistret. Accessed 10 January 2024. https://www.socialstyrelsen.se/globalassets/sharepoint‐dokument/artikelkatalog/statistik/2022‐2‐7767.pdf
- 20. Socialstyrelsen. National Prescribed Drug Register. Accessed 10 January 2024. https://www.socialstyrelsen.se/en/statistics‐and‐data/registers/national‐prescribed‐drug‐register/
- 21. Socialstyrelsen. Kvalitetsdeklaration statistik om läkemedel 2019. Accessed 10 January 2024. https://www.socialstyrelsen.se/globalassets/sharepoint‐dokument/artikelkatalog/statistik/2020‐4‐6707‐kvalitetsdeklaration.pdf
- 22. Socialstyrelsen. Det statistiska registrets framställning och kvalitet—Dödsorsaksregistret. Accessed 10 January 2024. https://www.socialstyrelsen.se/globalassets/sharepoint‐dokument/artikelkatalog/statistik/2022‐5‐7909.pdf
- 23. Miettinen O, Nurminen M. Comparative analysis of two rates. Stat Med. 1985;4:213‐226. [DOI] [PubMed] [Google Scholar]
- 24. Ladha KS, Wijeysundera DN. Role of patient‐centred outcomes after hospital discharge: a state‐of‐the‐art review. Anaesthesia. 2020;75(Suppl 1):e151‐e157. [DOI] [PubMed] [Google Scholar]
- 25. Merkow RP, Bilimoria KY, McCarter MD, et al. Post‐discharge venous thromboembolism after cancer surgery: extending the case for extended prophylaxis. Ann Surg. 2011;254:131‐137. [DOI] [PubMed] [Google Scholar]
- 26. Iyer R, Gentry‐Maharaj A, Nordin A, et al. Patient‐reporting improves estimates of postoperative complication rates: a prospective cohort study in gynaecological oncology. Br J Cancer. 2013;109:623‐632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Woodfield J, Deo P, Davidson A, Chen TY, van Rij A. Patient reporting of complications after surgery: what impact does documenting postoperative problems from the perspective of the patient using telephone interview and postal questionnaires have on the identification of complications after surgery? BMJ Open. 2019;9:e028561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Brandsborg B, Nikolajsen L. Chronic pain after hysterectomy. Curr Opin Anaesthesiol. 2018;31:268‐273. [DOI] [PubMed] [Google Scholar]
- 29. The COMET initiative. Accessed 17 January 2024. https://www.comet‐initiative.org/
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1.
Table S2.
Table S3.